cvpr cvpr2013 cvpr2013-59 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Mihir Jain, Hervé Jégou, Patrick Bouthemy
Abstract: Several recent works on action recognition have attested the importance of explicitly integrating motion characteristics in the video description. This paper establishes that adequately decomposing visual motion into dominant and residual motions, both in the extraction of the space-time trajectories and for the computation of descriptors, significantly improves action recognition algorithms. Then, we design a new motion descriptor, the DCS descriptor, based on differential motion scalar quantities, divergence, curl and shear features. It captures additional information on the local motion patterns enhancing results. Finally, applying the recent VLAD coding technique proposed in image retrieval provides a substantial improvement for action recognition. Our three contributions are complementary and lead to outperform all reported results by a significant margin on three challenging datasets, namely Hollywood 2, HMDB51 and Olympic Sports. 1. Introduction and related work Human actions often convey the essential meaningful content in videos. Yet, recognizing human actions in un- constrained videos is a challenging problem in Computer Vision which receives a sustained attention due to the potential applications. In particular, there is a large interest in designing video-surveillance systems, providing some automatic annotation of video archives as well as improving human-computer interaction. The solutions proposed to address this problem inherit, to a large extent, from the techniques first designed for the goal of image search and classification. The successful local features developed to describe image patches [15, 23] have been translated in the 2D+t domain as spatio-temporal local descriptors [13, 30] and now include motion clues [29]. These descriptors are often extracted from spatial-temporal interest points [12, 3 1]. More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32]. Most of these approaches produce large set of local descriptors which are in turn aggregated to produce a single vector representing the video, in order to enable the use of powerful discriminative classifiers such as support vector machines (SVMs). This is usually done with the bag- Figure 1. Optical flow field vectors (green vectors with red end points) before and after dominant motion compensation. Most of the flow vectors due to camera motion are suppressed after compensation. One of the contributions of this paper is to show that compensating for the dominant motion is beneficial for most of the existing descriptors used for action recognition. of-words technique [24], which quantizes the local features using a k-means codebook. Thanks to the successful combination of this encoding technique with the aforementioned local descriptors, the state of the art in action recognition is able to go beyond the toy problems ofclassifying simple human actions in controlled environment and considers the detection of actions in real movies or video clips [11, 16]. Despite these progresses, the existing descriptors suffer from an uncompleted handling of motion in the video sequence. Motion is arguably the most reliable source of information for action recognition, as often related to the actions of interest. However, it inevitably involves the background or camera motion when dealing with uncontrolled and re- alistic situations. Although some attempts have been made to compensate camera motion in several ways [10, 21, 26, 29, 32], how to separate action motion from that caused by the camera, and how to reflect it in the video description remains an open issue. The motion compensation mechanism employed in [10] is tailor-made to the Motion Interchange Pattern encoding technique. The Motion Boundary Histogram (MBH) [29] is a recent appealing approach to 222555555533 suppress the constant motion by considering the flow gradient. It is robust to some extent to the presence of camera motion, yet it does not explicitly handle the camera motion. Another approach [26] uses a sophisticated and robust (RANSAC) estimation of camera motion. It first segments the color image into regions corresponding to planar parts in the scene and estimates the (three) dominant homographies to update the motion associated with local features. A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level. All these works support the potential of motion compensation. As the first contribution of this paper, we address the problem in a way that departs from these works by considering the compensation of the dominant motion in both the tracking stages and encoding stages involved in the computation of action recognition descriptors. We rely on the pioneering works on motion compensation such as the technique proposed in [20], that considers 2D polynomial affine motion models for estimating the dominant image motion. We consider this particular model for its robustness and its low computational cost. It was already used in [21] to separate the dominant motion (assumed to be due to the camera motion) and the residual motion (corresponding to the independent scene motions) for dynamic event recognition in videos. However, the statistical modeling of both motion components was global (over the entire image) and only the normal flow was computed for the latter. Figure 1 shows the vectors of optical flow before and after applying the proposed motion compensation. Our method successfully suppresses most of the background motion and reinforces the focus towards the action of interest. We exploit this compensated motion both for descriptor computation and for extracting trajectories. However, we also show that the camera motion should not be thrown as it contains complementary information that is worth using to recognize certain action categories. Then, we introduce the Divergence-Curl-Shear (DCS) descriptor, which encodes scalar first-order motion features, namely the motion divergence, curl and shear. It captures physical properties of the flow pattern that are not involved in the best existing descriptors for action recognition, except in the work of [1] which exploits divergence and vorticity among a set of eleven kinematic features computed from the optical flow. Our DCS descriptor provides a good performance recognition performance on its own. Most importantly, it conveys some information which is not captured by existing descriptors and further improves the recognition performance when combined with the other descriptors. As a last contribution, we bring an encoding technique known as VLAD (vector oflocal aggregated descriptors) [8] to the field of action recognition. This technique is shown to be better than the bag-of-words representation for combining all the local video descriptors we have considered. The organization of the paper is as follows. Section 2 introduces the motion properties that we will consider through this paper. Section 3 presents the datasets and classification scheme used in our different evaluations. Section 4 details how we revisit several popular descriptors of the literature by the means of dominant motion compensation. Our DCS descriptor based on kinematic properties is introduced in Section 5 and improved by the VLAD encoding technique, which is introduced and bench-marked in Section 6 for several video descriptors. Section 7 provides a comparison showing the large improvement achieved over the state of the art. Finally, Section 8 concludes the paper. 2. Motion Separation and Kinematic Features In this section, we describe the motion clues we incorporate in our action recognition framework. We separate the dominant motion and the residual motion. In most cases, this will account to distinguishing the impact of camera movement and independent actions. Note that we do not aim at recovering the 3D camera motion: The 2D parametric motion model describes the global (or dominant) motion between successive frames. We first explain how we estimate the dominant motion and employ it to separate the dominant flow from the optical flow. Then, we will introduce kinematic features, namely divergence, curl and shear for a more comprehensive description of the visual motion. 2.1. Affine motion for compensating camera motion Among polynomial motion models, we consider the 2D affine motion model. Simplest motion models such as the 4parameter model formed by the combination of 2D translation, 2D rotation and scaling, or more complex ones such as the 8-parameter quadratic model (equivalent to a homography), could be selected as well. The affine model is a good trade-off between accuracy and efficiency which is of primary importance when processing a huge video database. It does have limitations since strictly speaking it implies a single plane assumption for the static background. However, this is not that penalizing (especially for outdoor scenes) if differences in depth remain moderated with respect to the distance to the camera. The affine flow vector at point p = (x, y) and at time t, is defined as waff(pt) =?cc12((t ) ?+?aa31((t ) aa42((t ) ? ?xytt?. (1) = + + = + uaff(pt) c1(t) a1(t)xt a2(t)yt and vaff(pt) c2(t) a3 (t)xt + a4(t)yt are horizontal and vertical components of waff(pt) respectively. Let us denote the optical flow vector at point p at time t as w(pt) = (u(pt) , v(pt)). We introduce the flow vector ω(pt) obtained by removing the affine flow vector from the optical flow vector ω(pt) = w(pt) − waff(pt) . (2) 222555555644 The dominant motion (estimated as waff(pt)) is usually due to the camera motion. In this case, Equation 2 amounts to canceling (or compensating) the camera motion. Note that this is not always true. For example in case of close-up on a moving actor, the dominant motion will be the affine estimation of the apparent actor motion. The interpretation of the motion compensation output will not be that straightforward in this case, however the resulting ω-field will still exhibit different patterns for the foreground action part and the background part. In the remainder, we will refer to the “compensated” flow as ω-flow. Figure 1 displays the computed optical flow and the ωflow. We compute the affine flow with the publicly available Motion2D software1 [20] which implements a realtime robust multiresolution incremental estimation framework. The affine motion model has correctly accounted for the motion induced by the camera movement which corresponds to the dominant motion in the image pair. Indeed, we observe that the compensated flow vectors in the background are close to null and the compensated flow in the foreground, i.e., corresponding to the actors, is conversely inflated. The experiments presented along this paper will show that effective separation of dominant motion from the residual motions is beneficial for action recognition. As explained in Section 4, we will compute local motion descriptors, such as HOF, on both the optical flow and the compensated flow (ω-flow), which allows us to explicitly and directly characterize the scene motion. 2.2. Local kinematic features By kinematic features, we mean local first-order differential scalar quantities computed on the flow field. We consider the divergence, the curl (or vorticity) and the hyperbolic terms. They inform on the physical pattern of the flow so that they convey useful information on actions in videos. They can be computed from the first-order derivatives of the flow at every point p at every frame t as ⎨⎪ ⎪ ⎪ ⎧hcdyuipvr1l2(p t) = −∂ u ∂ (yxp(xtp) +−∂ v ∂ v(px ypxt ) The diverg⎪⎩ence is related to axial motion, expansion scaling effects, the curl to rotation in the image plane. hyperbolic terms express the shear of the visual flow responding to more complex configuration. We take account the shear quantity only: shear(pt) = ?hyp12(pt) + hyp22(pt). (3) and The corinto (4) 1http://www.irisa.fr/vista/Motion2D/ In Section 5, we propose the DCS descriptor that is based on the kinematic features (divergence, curl and shear) of the visual motion discussed in this subsection. It is computed on either the optical or the compensated flow, ω-flow. 3. Datasets and evaluation This section first introduces the datasets used for the evaluation. Then, we briefly present the bag-of-feature model and the classification scheme used to encode the descriptors which will be introduced in Section 4. Hollywood2. The Hollywood2 dataset [16] contains 1,707 video clips from 69 movies representing 12 action classes. It is divided into train set and test set of 823 and 884 samples respectively. Following the standard evaluation protocol of this benchmark, we use average precision (AP) for each class and the mean of APs (mAP) for evaluation. HMDB51. The HMDB51 dataset [11] is a large dataset containing 6,766 video clips extracted from various sources, ranging from movies to YouTube. It consists of 51 action classes, each having at least 101 samples. We follow the evaluation protocol of [11] and use three train/test splits, each with 70 training and 30 testing samples per class. The average classification accuracy is computed over all classes. Out of the two released sets, we use the original set as it is more challenging and used by most of the works reporting results in action recognition. Olympic Sports. The third dataset we use is Olympic Sports [19], which again is obtained from YouTube. This dataset contains 783 samples with 16 sports action classes. We use the provided2 train/test split, there are 17 to 56 training samples and 4 to 11test samples per class. Mean AP is used for the evaluation, which is the standard choice. Bag of features and classification setup. We first adopt the standard BOF [24] approach to encode all kinds of descriptors. It produces a vector that serves as the video representation. The codebook is constructed for each type of descriptor separately by the k-means algorithm. Following a common practice in the literature [27, 29, 30], the codebook size is set to k=4,000 elements. Note that Section 6 will consider encoding technique for descriptors. For the classification, we use a non-linear SVM with χ2kernel. When combining different descriptors, we simply add the kernel matrices, as done in [27]: K(xi,xj) = exp?−?cγ1cD(xic,xjc)?, 2http://vision.stanford.edu/Datasets/OlympicSports/ 222555555755 (5) where D(xic, xjc) is χ2 distance between video xic and xjc with respect to c-th channel, corresponding to c-th descriptor. The quantity γc is the mean value of χ2 distances between the training samples for the c-th channel. The multiclass classification problem that we consider is addressed by applying a one-against-rest approach. 4. Compensated descriptors This section describes how the compensation ofthe dominant motion is exploited to improve the quality of descriptors encoding the motion and the appearance around spatio-temporal positions, hence the term “compensated descriptors”. First, we briefly review the local descriptors [5, 13, 16, 29, 30] used here along with dense trajectories [29]. Second, we analyze the impact of motion flow compensation when used in two different stages of the descriptor computation, namely in the tracking and the description part. 4.1. Dense trajectories and local descriptors Employing dense trajectories to compute local descriptors is one of the state-of-the-art approaches for action recognition. It has been shown [29] that when local descriptors are computed over dense trajectories the performance improves considerably compared to when computed over spatio temporal features [30]. Dense Trajectories [29]: The trajectories are obtained by densely tracking sampled points using optical flow fields. First, feature points are sampled from a dense grid, with step size of 5 pixels and over 8 scales. Each feature point pt = (xt, yt) at frame t is then tracked to the next frame by median filtering in a dense optical flow field F = (ut, vt) as follows: pt+1 = (xt+1 , yt+1) = (xt, yt) + (M ∗ F) | (x ¯t,y ¯t) , (6) where M is the kernel of median filtering and ( x¯ t, y¯ t) is the rounded position of (xt, yt). The tracking is limited to L (=15) frames to avoid any drifting effect. Excessively short trajectories and trajectories exhibiting sudden large displacements are removed as they induce some artifacts. Trajectories must be understood here as tracks in the spacetime volume of the video. Local descriptors: The descriptors are computed within a space-time volume centered around each trajectory. Four types of descriptors are computed to encode the shape of the trajectory, local motion pattern and appearance, namely Trajectory [29], HOF (histograms of optical flow) [13], MBH [4] and HOG (histograms of oriented gradients) [3]. All these descriptors depend on the flow field used for the tracking and as input of the descriptor computation: 1. The Trajectory descriptor encodes the shape of the trajectory represented by the normalized relative coor- × dinates of the successive points forming the trajectory. It directly depends on the dense flow used for tracking points. 2. HOF is computed using the orientations and magnitudes of the flow field. 3. MBH is designed to capture the gradient of horizontal and vertical components of the flow. The motion boundaries encode the relative pixel motion and therefore suppress camera motion, but only to some extent. 4. HOG encodes the appearance by using the intensity gradient orientations and magnitudes. It is formally not a motion descriptor. Yet the position where the descriptor is computed depends on the trajectory shape. As in [29], volume around a feature point is divided into a 2 2 3 space-time grid. The orientations are quantized ian 2to × ×8 b2i ×ns 3fo srp HacOe-Gti amned g g9r ibdi.ns T fhoer o oHriOenFt (awtioitnhs one a qdudainttiiozneadl zero bin). The horizontal and vertical components of MBH are separately quantized into 8 bins each. 4.2. Impact of motion compensation The optical flow is simply referred to as flow in the following, while the compensated flow (see subsection 2. 1) is denoted by ω-flow. Both of them are considered in the tracking and descriptor computation stages. The trajectories obtained by tracking with the ω-flow are called ω-trajectories. Figure 2 comparatively illustrates the ωtrajectories and the trajectories obtained using the flow. The input video shows a man moving away from the car. In this video excerpt, the camera is following the man walking to the right, thus inducing a global motion to the left in the video. When using the flow, the computed trajectories reflect the combination of these two motion components (camera and scene motion) as depicted by Subfigure 2(b), which hampers the characterization of the current action. In contrast, the ω-trajectories plotted in Subfigure 2(c) are more active on the actor moving on the foreground, while those localized in the background are now parallel to the time axis enhancing static parts of the scene. The ω-trajectories are therefore more relevant for action recognition, since they are more regularly and more exclusively following the actor’s motion. Impact on Trajectory and HOG descriptors. Table 1reports the impact of ω-trajectories on Trajectory and HOG descriptors, which are both significantly improved by 3%4% of mAP on the two datasets. When improved by ωflow, these descriptors will be respectively referred to as ω-Trajdesc and ω-HOG in the rest of the paper. Although the better performance of ω-Trajdesc versus the original Trajectory descriptor was expected, the one 222555555866 2. Trajectories obtained from optical and compensated flows. The green tail is the trajectory the current frame. The trajectories are sub-sampled for the sake of clarity. The frames are extracted Figure over every 15 frames with red dot indicating 5 frames in this example. DescriptorHollywood2HMDB51 BaseTrliaωnje- Tc(rtoarejrdpyreos[c2d9u]ced)54 7 1. 7 4% %2382.–89% BaseliHnωOe- (GHreOp [2rG9od]uced)4 451 . 658%%%2296.– 13%% Table 1. ω-Trajdesc and ω-HOG: Impact of compensating flow on Trajectory descriptor and HOG descriptors. achieved by ω-HOG might be surprising. Our interpretation is that HOG captures more context with the modified trajectories. More precisely, the original HOG descriptor is computed from a 2D+t sub-volume aligned with the corresponding trajectory and hence represents the appearance along the trajectory shape. When using ω-flow, we do not align the video sequence. As a result, the ω-HOG descriptor is no more computed around the very same tracked physical point in the space-time volume but around points lying in a patch of the initial feature point, whose size depends on the affine flow magnitude. ω-HOG can be viewed as a “patchbased” computation capturing more information about the appearance of the background or of the moving foreground. As for ω-trajectories, they are closer to the real trajectories of the moving actors as they usually cancel the camera movement, and so, more easier to train and recognize. Impact on HOF. The ω-flow impacts computation used as an input to HOF computation itself. Therefore, HOF can both types of trajectories (ω-trajectories both the trajectory and the descriptor be computed along or those extracted MethodHollywood2HMDB51 Table(ω2rHf.-alocO IwomkF)inpHgacOtFobf[2u9ωhsb]i:f-nlo ωgwot-hwωHOflFown5H 02 34O. 58291F% %descripto3 r706s38.:–1076% m%APfor Hollywood2 and average accuracy for HMDB5 1. The ω-HOF is used in subsequent evaluations. from flow) and can encode both kinds of flows (ω-flow or flow). For the sake of completeness, we evaluate all the variants as well as the combination of both flows in the descriptor computation stage. The results are presented in Table 2 and demonstrate the significant improvement obtained by computing the HOF descriptor with the ω-flow instead of the optical flow. Note that the type of trajectories which is used, either “Tracking flow” or “Tracking ω-flow”, has a limited impact in this case. From now on, we only consider the “Tracking ω-flow” case where HOF is computed along ω-trajectories. Interestingly, combining the HOF computed from the flow and the ω-flow further improves the results. This suggests that the two flow fields are complementary and the affine flow that was subtracted from ω-flow brings in additional information. For the sake of brevity, the combination of the two kinds of HOF, i.e., computed from the flow and the ω-flow using ω-trajectories, is referred to as the ω-HOF 222555555977 MethodHollywood2HMDB51 Tab(lerT3a.cIkmM inpBgMacHgωtBf-loH w [u2)s9in]gω f-lo wo MBH5 d42 e.052s7c% riptos:m34A90P.–3769f% orHllywood2 and average accuracy for HMDB5 1. DTerHasMjcBrOeblitpHGeFor4.ySumTωraw- frcilykto hw ionfgtheduωpCs-fcaolrtωmeiwNp-df/tl+Aωoutrw-finlogwthdesωcr- isTpc-fHtrMloaiOjrBpdswtGeHFosrc descriptor in the rest of this paper. Compared to the HOF baseline, the ω-HOF descriptor achieves a gain of +3.1% of mAP on Hollywood 2 and of +7.8% on HMDB51. Impact on MBH. Since MBH is computed from gradient of flow and cancel the constant motion, there is practically no benefit in using the ω-flow to compute the MBH descriptors, as shown in Table 3. However, by tracking ω-flow, the performance improves by around 1.3% for HMDB5 1 dataset and drops by around 1.5% for Hollywood2. This relative performance depends on the encoding technique. We will come back on this descriptor when considering another encoding scheme for local descriptors in Section 6. 4.3. Summary of compensated descriptors Table 4 summarizes the refined versions of the descriptors obtained by exploiting the ω-flow, and both ω-flow and the optical flow in the case of HOF. The revisited descriptors considerably improve the results compared to the orig- inal ones, with the noticeable exception of ω-MBH which gives mixed performance with a bag-of-features encoding scheme. But we already mention as this point that this incongruous behavior of ω-MBH is stabilized with the VLAD encoding scheme considered in Section 6. Another advantage of tracking the compensated flow is that fewer trajectories are produced. For instance, the total number of trajectories decreases by about 9. 16% and 22.81% on the Hollywood2 and HMDB51 datasets, respectively. Note that exploiting both the flow and the ω-flow do not induce much computational overhead, as the latter is obtained from the flow and the affine flow which is computed in real-time and already used to get the ω-trajectories. The only additional computational cost that we introduce by using the descriptors summarized in Table 4 is the computation of a second HOF descriptor, but this stage is relatively efficient and not the bottleneck of the extraction procedure. 5. Divergence-Curl-Shear descriptor This section introduces a new descriptor encoding the kinematic properties of motion discussed in Section 2.2. It is denoted by DCS in the rest of this paper. Combining kinematic features. The spatial derivatives are computed for the horizontal and vertical components of the flow field, which are used in turn to compute the divergence, curl and shear scalar values, see Equation 3. We consider all possible pairs of kinematic features, namely (div, curl), (div, shear) and (curl, shear). At each × ×× pixel, we compute the orientation and magnitude of the 2-D vector corresponding to each of these pairs. The orientation is quantized into histograms and the magnitude is used for weighting, similar to SIFT. Our motivation for encoding pairs is that the joint distribution of kinematic features conveys more information than exploiting them independently. Implementation details. The descriptor computation and parameters are similar to HOG and other popular descriptors such as MBH, HOF. We obtain 8-bin histograms for each of the three feature pairs or components of DCS. The range of possible angles is 2π for the (div,curl) pair and π for the other pairs, because the shear is always positive. The DCS descriptor is computed for a space-time volume aligned with a trajectory, as done with the four descriptors mentioned in the previous section. In order to capture the spatio-temporal structure of kinematic features, the volume (32 32 pixels and L = 15 frames) is subdivided into a spatio-temporal grid nofd s Lize = nx 5× f ny m×e nt, sw situhb nx =de ny =to 2a and nt = 3. These parameters ×hnave× × bneen fixed for the sake of consistency with the other descriptors. For each pair of kinematic features, each cell in the grid is represented by a histogram. The resulting local descriptors have a dimensionality equal to 288 = nx ny nt 8 3. At the video level, these descriptors are nenc×od end i×nto 8 a single vector representation using either BOF or the VLAD encoding scheme introduced in the next section. 6. VLAD in actions VLAD [8] is a descriptor encoding technique that aggregates the descriptors based on a locality criterion in the feature space. To our knowledge, this technique has never been considered for action recognition. Below, we briefly introduce this approach and give the performance achieved for all the descriptors introduced along the previous sections. VLAD in brief. Similar to BOF, VLAD relies on a codebook C = {c1, c2 , ...ck} of k centroids learned by k-means. bTohoek representation is ob}t oaifn ked c by summing, efodr b yea kch-m mveiasunasl. word ci, the differences x − ci of the vectors x assigned to ci, thereby producing a sv exct −or c representation oflength d×k, 222555556088 DMeBscHriptorV5 LH.A1o%Dlywo5Bo4d.O2 %F4V3L.3HA%MD B35B91.O7%F Taωbl-eDHM5rOCBa.FSjGPdHe+rsωfco-mMHaBOnFeofV54L2936A.51D% with5431ω208-.5T96% rajde3s42c97158,.ω3% -HOG342,58019ω.6-% HOF descriptors and their combination. where d is the dimension ofthe local descriptors. We use the codebook size, k = 256. Despite this large dimensionality, VLAD is efficient because it is effectively compared with a linear kernel. VLAD is post-processed using a componentwise power normalization, which dramatically improves its performance [8]. While cross validating the parameter α involved in this power normalization, we consistently observe, for all the descriptors, a value between 0.15 and 0.3. Therefore, this parameter is set to α = 0.2 in all our experiments. For classification, we use a linear SVM and oneagainst-rest approach everywhere, unless stated otherwise. Impact on existing descriptors. We employ VLAD because it is less sensitive to quantization parameters and appears to provide better performance with descriptors having a large dimensionality. These properties are interesting in our case, because the quantization parameters involved in the DCS and MBH descriptors have been used unchanged in Section 4 for the sake of direct comparison. They might be suboptimal when using the ω-flow instead of the optical flow on which they have initially been optimized [29]. Results for MBH and ω-MBH in Table 5 supports this argument. When using VLAD instead of BOF, the scores are stable in both the cases and there is no mixed inference as that observed in Table 3. VLAD also has significant positive influence on accuracy of ω-DCS descriptor. We also observe that ω-DCS is complementary to ω-MBH and adds to the performance. Still DCS is probably not best utilized in the current setting of parameters. In case of ω-Trajdesc and ω-HOG, the scores are better with BOF on both the datasets. ω-HOF with VLAD improves on HMDB5 1, but remains equivalent for Hollywood2. Although BOF leads to better scores for the descriptors considered individually, their combination with VLAD outperforms the BOF. 7. Comparison with the state of the art This section reports our results with all descriptors combined and compares our method with the state of the art. TrajectorCy+omHbOiGna+tHioOnF+ MDBCHSHol5 l98y.w76%o%od2H4M489.D02%B%51 All ω-descriptors all five compensated descriptors using combined62.5%52.1% Table 6. Combination of VLAD representation. WU*JVliaOnughreM tHaeolth. [yo2w9d87o] 256 0985. 37% SKa*duOeJinhrau tnegdMteHatlMa.h [ol1Dd.0B[91]25 24 609.8172% Table 7. Comparison with the state of the art on Hollywood2 and HMDB5 1 datasets. *Vig et al. [28] gets 61.9% by using external eye movements data. *Jiang et al. [9] used one-vs-one multi class SVM while our and other methods use one-vs-rest SVMs. With one-against-one multi class SVM we obtain 45. 1% for HMDB51. Descriptor combination. Table 6 reports the results obtained when the descriptors are combined. Since we use VLAD, our baseline is updated that is combination of Trajectory, HOG, HOF and MBH with VLAD representation. When DCS is added to the baseline there is an improvement of 0.9% and 1.2%. With combination of all five compensated descriptors we obtain 62.5% and 52.1% on the two datasets. This is a large improvement even over the updated baseline, which shows that the proposed motion compensation and the way we exploit it are significantly important for action recognition. The comparison with the state of the art is shown in Table 7. Our method outperforms all the previously reported results in the literature. In particular, on the HMDB51 dataset, the improvement over the best reported results to date is more than 11% in average accuracy. Jiang el al. [9] used a one-against-one multi-class SVM, which might have resulted in inferior scores. With a similar multi-class SVM approach, our method obtains 45. 1%, which remains significantly better than their result. All others results were reported with one-against-rest approach. On Olympic Sports dataset we obtain mAP of 83.2% with ‘All ω-descriptors combined’ and the improvement is mostly because of VLAD and ω-flow. The best reported mAPs on this dataset are Liu et al. [14] (74.4%) and Jiang et al. [9] (80.6%), which we exceed convincingly. Gaidon et al. [6] reports the best average accuracy of 82.7%. 8. Conclusions This paper first demonstrates the interest of canceling the dominant motion (predominantly camera motion) to make the visual motion truly related to actions, for both the trajectory extraction and descriptor computation stages. It pro222555556199 duces significantly better versions (called compensated descriptors) of several state-of-the-art local descriptors for action recognition. The simplicity, efficiency and effectiveness of this motion compensation approach make it applicable to any action recognition framework based on motion descriptors and trajectories. The second contribution is the new DCS descriptor derived from the first-order scalar motion quantities specifying the local motion patterns. It captures additional information which is proved complementary to the other descriptors. Finally, we show that VLAD encoding technique instead of bag-of-words boosts several action descriptors, and overall exhibits a significantly better performance when combining different types of descriptors. Our contributions are all complementary and significantly outperform the state of the art when combined, as demon- strated by our extensive experiments on the Hollywood 2, HMDB51 and Olympic Sports datasets. Acknowledgments This work was supported by the Quaero project, funded by Oseo, French agency for innovation. We acknowledge Heng Wang’s help for reproducing some of their results. References [1] S. Ali and M. Shah. Human action recognition in videos using kinematic features and multiple instance learning. IEEE T-PAMI, 32(2):288–303, Feb. 2010. [2] T. Brox and J. Malik. Object segmentation by long term analysis of point trajectories. In ECCV, Sep. 2010. [3] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In CVPR, Jun. 2005. [4] N. Dalal, B. Triggs, and C. Schmid. Human detection using oriented histograms of flow and appearance. In ECCV, May 2006. [5] P. Dollar, V. Rabaud, G. Cottrell, and S. Belongie. Behavior recognition via sparse spatio-temporal features. In VS-PETS, Oct. 2005. [6] A. Gaidon, Z. Harchaoui, and C. Schmid. Recognizing activities with cluster-trees of tracklets. In BMVC, Sep. 2012. [7] A. Hervieu, P. Bouthemy, and J.-P. Le Cadre. A statistical video content recognition method using invariant features on object trajectories. IEEE T-CSVT, 18(1 1): 1533–1543, 2008. [8] H. J ´egou, F. Perronnin, M. Douze, J. S ´anchez, P. P ´erez, and C. Schmid. Aggregating local descriptors into compact [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] codes. IEEE T-PAMI, 34(9):1704–1716, 2012. Y.-G. Jiang, Q. Dai, X. Xue, W. Liu, and C.-W. Ngo. Trajectory-based modeling of human actions with motion reference points. In ECCV, Oct. 2012. O. Kliper-Gross, Y. Gurovich, T. Hassner, and L. Wolf. Motion interchange patterns for action recognition in unconstrained videos. In ECCV, Oct. 2012. H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. Hmdb: A large video database for human motion recognition. In ICCV, Nov. 2011. I. Laptev and T. Lindeberg. Space-time interest points. In ICCV, Oct. 2003. I. Laptev, M. Marzalek, C. Schmid, and B. Rozenfeld. Learning realistic human actions from movies. In CVPR, Jun. 2008. J. Liu, B. Kuipers, and S. Savarese. Recognizing human actions by attributes. In CVPR, Jun. 2011. D. Lowe. Distinctive image features from scale-invariant keypoints. IJCV, 60(2):91–1 10, Nov. 2004. M. Marzalek, I. Laptev, and C. Schmid. Actions in context. In CVPR, Jun. 2009. P. Matikainen, M. Hebert, and R. Sukthankar. Trajectons: Action recognition through the motion analysis of tracked features. In Workshop on Video-Oriented Object and Event Classification, ICCV, Sep. 2009. R. Messing, C. J. Pal, and H. A. Kautz. Activity recognition using the velocity histories of tracked keypoints. In ICCV, Sep. 2009. J. C. Niebles, C.-W. Chen, and F.-F. Li. Modeling temporal structure of decomposable motion segments for activity classification. In ECCV, Sep. 2010. [20] J.-M. Odobez and P. Bouthemy. Robust multiresolution estimation of parametric motion models. Jal of Vis. Comm. and Image Representation, 6(4):348–365, Dec. 1995. [21] G. Piriou, P. Bouthemy, and J.-F. Yao. Recognition of dynamic video contents with global probabilistic models of visual motion. IEEE T-IP, 15(1 1):3417–3430, 2006. [22] S. Sadanand and J. J. Corso. Action bank: A high-level representation of activity in video. In CVPR, Jun. 2012. [23] C. Schmid and R. Mohr. Local grayvalue invariants for image retrieval. IEEE T-PAMI, 19(5):530–534, May 1997. [24] J. Sivic and A. Zisserman. Video Google: A text retrieval approach to object matching in videos. In ICCV, pages 1470– 1477, Oct. 2003. [25] J. Sun, X. Wu, S. Yan, L. F. Cheong, T.-S. Chua, and J. Li. Hierarchical spatio-temporal context modeling for action recognition. In CVPR, Jun. 2009. [26] H. Uemura, S. Ishikawa, and K. Mikolajczyk. Feature tracking and motion compensation for action recognition. In BMVC, Sep. 2008. [27] M. M. Ullah, S. N. Parizi, and I. Laptev. Improving bag-offeatures action recognition with non-local cues. In BMVC, Sep. 2010. [28] E. Vig, M. Dorr, and D. Cox. Saliency-based space-variant descriptor sampling for action recognition. In ECCV, Oct. 2012. [29] H. Wang, A. Kl¨ aser, C. Schmid, and C.-L. Liu. Action recognition by dense trajectories. In CVPR, Jun. 2011. [30] H. Wang, M. M. Ullah, A. Kl¨ aser, I. Laptev, and C. Schmid. Evaluation of local spatio-temporal features for action recognition. In BMVC, Sep. 2009. [3 1] G. Willems, T. Tuytelaars, and L. J. V. Gool. An efficient dense and scale-invariant spatio-temporal interest point detector. In ECCV, Oct. 2008. [32] S. Wu, O. Oreifej, and M. Shah. Action recognition in videos acquired by a moving camera using motion decomposition of lagrangian particle trajectories. In ICCV, Nov. 2011. 222555666200
Reference: text
sentIndex sentText sentNum sentScore
1 Better exploiting motion for better action recognition Mihir Jain Herv e´ J e´gou Patrick Bouthemy INRIA, Rennes, France Abstract Several recent works on action recognition have attested the importance of explicitly integrating motion characteristics in the video description. [sent-1, score-1.137]
2 This paper establishes that adequately decomposing visual motion into dominant and residual motions, both in the extraction of the space-time trajectories and for the computation of descriptors, significantly improves action recognition algorithms. [sent-2, score-0.946]
3 Then, we design a new motion descriptor, the DCS descriptor, based on differential motion scalar quantities, divergence, curl and shear features. [sent-3, score-0.916]
4 The successful local features developed to describe image patches [15, 23] have been translated in the 2D+t domain as spatio-temporal local descriptors [13, 30] and now include motion clues [29]. [sent-12, score-0.493]
5 More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32]. [sent-14, score-0.443]
6 Optical flow field vectors (green vectors with red end points) before and after dominant motion compensation. [sent-17, score-0.674]
7 Most of the flow vectors due to camera motion are suppressed after compensation. [sent-18, score-0.609]
8 One of the contributions of this paper is to show that compensating for the dominant motion is beneficial for most of the existing descriptors used for action recognition. [sent-19, score-0.911]
9 Despite these progresses, the existing descriptors suffer from an uncompleted handling of motion in the video sequence. [sent-22, score-0.521]
10 Motion is arguably the most reliable source of information for action recognition, as often related to the actions of interest. [sent-23, score-0.335]
11 However, it inevitably involves the background or camera motion when dealing with uncontrolled and re- alistic situations. [sent-24, score-0.338]
12 Although some attempts have been made to compensate camera motion in several ways [10, 21, 26, 29, 32], how to separate action motion from that caused by the camera, and how to reflect it in the video description remains an open issue. [sent-25, score-0.91]
13 The motion compensation mechanism employed in [10] is tailor-made to the Motion Interchange Pattern encoding technique. [sent-26, score-0.509]
14 The Motion Boundary Histogram (MBH) [29] is a recent appealing approach to 222555555533 suppress the constant motion by considering the flow gradient. [sent-27, score-0.538]
15 It first segments the color image into regions corresponding to planar parts in the scene and estimates the (three) dominant homographies to update the motion associated with local features. [sent-30, score-0.403]
16 A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level. [sent-31, score-0.421]
17 As the first contribution of this paper, we address the problem in a way that departs from these works by considering the compensation of the dominant motion in both the tracking stages and encoding stages involved in the computation of action recognition descriptors. [sent-33, score-1.085]
18 We rely on the pioneering works on motion compensation such as the technique proposed in [20], that considers 2D polynomial affine motion models for estimating the dominant image motion. [sent-34, score-0.949]
19 It was already used in [21] to separate the dominant motion (assumed to be due to the camera motion) and the residual motion (corresponding to the independent scene motions) for dynamic event recognition in videos. [sent-36, score-0.799]
20 However, the statistical modeling of both motion components was global (over the entire image) and only the normal flow was computed for the latter. [sent-37, score-0.593]
21 Figure 1 shows the vectors of optical flow before and after applying the proposed motion compensation. [sent-38, score-0.628]
22 Our method successfully suppresses most of the background motion and reinforces the focus towards the action of interest. [sent-39, score-0.517]
23 We exploit this compensated motion both for descriptor computation and for extracting trajectories. [sent-40, score-0.654]
24 However, we also show that the camera motion should not be thrown as it contains complementary information that is worth using to recognize certain action categories. [sent-41, score-0.629]
25 Then, we introduce the Divergence-Curl-Shear (DCS) descriptor, which encodes scalar first-order motion features, namely the motion divergence, curl and shear. [sent-42, score-0.783]
26 It captures physical properties of the flow pattern that are not involved in the best existing descriptors for action recognition, except in the work of [1] which exploits divergence and vorticity among a set of eleven kinematic features computed from the optical flow. [sent-43, score-1.17]
27 As a last contribution, we bring an encoding technique known as VLAD (vector oflocal aggregated descriptors) [8] to the field of action recognition. [sent-46, score-0.38]
28 Section 4 details how we revisit several popular descriptors of the literature by the means of dominant motion compensation. [sent-51, score-0.602]
29 Our DCS descriptor based on kinematic properties is introduced in Section 5 and improved by the VLAD encoding technique, which is introduced and bench-marked in Section 6 for several video descriptors. [sent-52, score-0.458]
30 Motion Separation and Kinematic Features In this section, we describe the motion clues we incorporate in our action recognition framework. [sent-56, score-0.568]
31 We separate the dominant motion and the residual motion. [sent-57, score-0.437]
32 Note that we do not aim at recovering the 3D camera motion: The 2D parametric motion model describes the global (or dominant) motion between successive frames. [sent-59, score-0.627]
33 We first explain how we estimate the dominant motion and employ it to separate the dominant flow from the optical flow. [sent-60, score-0.9]
34 Then, we will introduce kinematic features, namely divergence, curl and shear for a more comprehensive description of the visual motion. [sent-61, score-0.542]
35 Affine motion for compensating camera motion Among polynomial motion models, we consider the 2D affine motion model. [sent-64, score-1.293]
36 The affine flow vector at point p = (x, y) and at time t, is defined as waff(pt) =? [sent-69, score-0.366]
37 Let us denote the optical flow vector at point p at time t as w(pt) = (u(pt) , v(pt)). [sent-77, score-0.361]
38 We introduce the flow vector ω(pt) obtained by removing the affine flow vector from the optical flow vector ω(pt) = w(pt) − waff(pt) . [sent-78, score-0.998]
39 (2) 222555555644 The dominant motion (estimated as waff(pt)) is usually due to the camera motion. [sent-79, score-0.474]
40 For example in case of close-up on a moving actor, the dominant motion will be the affine estimation of the apparent actor motion. [sent-82, score-0.58]
41 The interpretation of the motion compensation output will not be that straightforward in this case, however the resulting ω-field will still exhibit different patterns for the foreground action part and the background part. [sent-83, score-0.665]
42 Figure 1 displays the computed optical flow and the ωflow. [sent-85, score-0.392]
43 We compute the affine flow with the publicly available Motion2D software1 [20] which implements a realtime robust multiresolution incremental estimation framework. [sent-86, score-0.389]
44 The affine motion model has correctly accounted for the motion induced by the camera movement which corresponds to the dominant motion in the image pair. [sent-87, score-1.129]
45 Indeed, we observe that the compensated flow vectors in the background are close to null and the compensated flow in the foreground, i. [sent-88, score-0.962]
46 The experiments presented along this paper will show that effective separation of dominant motion from the residual motions is beneficial for action recognition. [sent-91, score-0.712]
47 As explained in Section 4, we will compute local motion descriptors, such as HOF, on both the optical flow and the compensated flow (ω-flow), which allows us to explicitly and directly characterize the scene motion. [sent-92, score-1.109]
48 Local kinematic features By kinematic features, we mean local first-order differential scalar quantities computed on the flow field. [sent-95, score-0.701]
49 They inform on the physical pattern of the flow so that they convey useful information on actions in videos. [sent-97, score-0.404]
50 They can be computed from the first-order derivatives of the flow at every point p at every frame t as ⎨⎪ ⎪ ⎪ ⎧hcdyuipvr1l2(p t) = −∂ u ∂ (yxp(xtp) +−∂ v ∂ v(px ypxt ) The diverg⎪⎩ence is related to axial motion, expansion scaling effects, the curl to rotation in the image plane. [sent-98, score-0.481]
51 hyperbolic terms express the shear of the visual flow responding to more complex configuration. [sent-99, score-0.468]
52 fr/vista/Motion2D/ In Section 5, we propose the DCS descriptor that is based on the kinematic features (divergence, curl and shear) of the visual motion discussed in this subsection. [sent-104, score-0.755]
53 It is computed on either the optical or the compensated flow, ω-flow. [sent-105, score-0.331]
54 The Hollywood2 dataset [16] contains 1,707 video clips from 69 movies representing 12 action classes. [sent-110, score-0.377]
55 Compensated descriptors This section describes how the compensation ofthe dominant motion is exploited to improve the quality of descriptors encoding the motion and the appearance around spatio-temporal positions, hence the term “compensated descriptors”. [sent-140, score-1.31]
56 First, we briefly review the local descriptors [5, 13, 16, 29, 30] used here along with dense trajectories [29]. [sent-141, score-0.408]
57 Second, we analyze the impact of motion flow compensation when used in two different stages of the descriptor computation, namely in the tracking and the description part. [sent-142, score-1.008]
58 Dense trajectories and local descriptors Employing dense trajectories to compute local descriptors is one of the state-of-the-art approaches for action recognition. [sent-145, score-1.033]
59 It has been shown [29] that when local descriptors are computed over dense trajectories the performance improves considerably compared to when computed over spatio temporal features [30]. [sent-146, score-0.495]
60 Dense Trajectories [29]: The trajectories are obtained by densely tracking sampled points using optical flow fields. [sent-147, score-0.595]
61 Excessively short trajectories and trajectories exhibiting sudden large displacements are removed as they induce some artifacts. [sent-151, score-0.352]
62 Four types of descriptors are computed to encode the shape of the trajectory, local motion pattern and appearance, namely Trajectory [29], HOF (histograms of optical flow) [13], MBH [4] and HOG (histograms of oriented gradients) [3]. [sent-154, score-0.645]
63 All these descriptors depend on the flow field used for the tracking and as input of the descriptor computation: 1. [sent-155, score-0.671]
64 It directly depends on the dense flow used for tracking points. [sent-157, score-0.362]
65 The motion boundaries encode the relative pixel motion and therefore suppress camera motion, but only to some extent. [sent-162, score-0.631]
66 Yet the position where the descriptor is computed depends on the trajectory shape. [sent-166, score-0.328]
67 Impact of motion compensation The optical flow is simply referred to as flow in the following, while the compensated flow (see subsection 2. [sent-173, score-1.528]
68 In this video excerpt, the camera is following the man walking to the right, thus inducing a global motion to the left in the video. [sent-179, score-0.393]
69 When using the flow, the computed trajectories reflect the combination of these two motion components (camera and scene motion) as depicted by Subfigure 2(b), which hampers the characterization of the current action. [sent-180, score-0.522]
70 ω-Trajdesc and ω-HOG: Impact of compensating flow on Trajectory descriptor and HOG descriptors. [sent-196, score-0.473]
71 More precisely, the original HOG descriptor is computed from a 2D+t sub-volume aligned with the corresponding trajectory and hence represents the appearance along the trajectory shape. [sent-199, score-0.482]
72 As a result, the ω-HOG descriptor is no more computed around the very same tracked physical point in the space-time volume but around points lying in a patch of the initial feature point, whose size depends on the affine flow magnitude. [sent-201, score-0.619]
73 As for ω-trajectories, they are closer to the real trajectories of the moving actors as they usually cancel the camera movement, and so, more easier to train and recognize. [sent-203, score-0.334]
74 Therefore, HOF can both types of trajectories (ω-trajectories both the trajectory and the descriptor be computed along or those extracted MethodHollywood2HMDB51 Table(ω2rHf. [sent-206, score-0.504]
75 Interestingly, combining the HOF computed from the flow and the ω-flow further improves the results. [sent-216, score-0.327]
76 This suggests that the two flow fields are complementary and the affine flow that was subtracted from ω-flow brings in additional information. [sent-217, score-0.678]
77 Since MBH is computed from gradient of flow and cancel the constant motion, there is practically no benefit in using the ω-flow to compute the MBH descriptors, as shown in Table 3. [sent-230, score-0.334]
78 We will come back on this descriptor when considering another encoding scheme for local descriptors in Section 6. [sent-235, score-0.436]
79 Summary of compensated descriptors Table 4 summarizes the refined versions of the descriptors obtained by exploiting the ω-flow, and both ω-flow and the optical flow in the case of HOF. [sent-238, score-0.969]
80 Another advantage of tracking the compensated flow is that fewer trajectories are produced. [sent-241, score-0.715]
81 Note that exploiting both the flow and the ω-flow do not induce much computational overhead, as the latter is obtained from the flow and the affine flow which is computed in real-time and already used to get the ω-trajectories. [sent-245, score-0.939]
82 Divergence-Curl-Shear descriptor This section introduces a new descriptor encoding the kinematic properties of motion discussed in Section 2. [sent-248, score-0.813]
83 The spatial derivatives are computed for the horizontal and vertical components of the flow field, which are used in turn to compute the divergence, curl and shear scalar values, see Equation 3. [sent-252, score-0.757]
84 The descriptor computation and parameters are similar to HOG and other popular descriptors such as MBH, HOF. [sent-258, score-0.376]
85 The DCS descriptor is computed for a space-time volume aligned with a trajectory, as done with the four descriptors mentioned in the previous section. [sent-261, score-0.398]
86 At the video level, these descriptors are nenc×od end i×nto 8 a single vector representation using either BOF or the VLAD encoding scheme introduced in the next section. [sent-266, score-0.348]
87 VLAD in actions VLAD [8] is a descriptor encoding technique that aggregates the descriptors based on a locality criterion in the feature space. [sent-268, score-0.557]
88 They might be suboptimal when using the ω-flow instead of the optical flow on which they have initially been optimized [29]. [sent-300, score-0.361]
89 D02%B%51 All ω-descriptors all five compensated descriptors using combined62. [sent-313, score-0.409]
90 With combination of all five compensated descriptors we obtain 62. [sent-337, score-0.433]
91 This is a large improvement even over the updated baseline, which shows that the proposed motion compensation and the way we exploit it are significantly important for action recognition. [sent-340, score-0.689]
92 Conclusions This paper first demonstrates the interest of canceling the dominant motion (predominantly camera motion) to make the visual motion truly related to actions, for both the trajectory extraction and descriptor computation stages. [sent-360, score-1.107]
93 It pro222555556199 duces significantly better versions (called compensated descriptors) of several state-of-the-art local descriptors for action recognition. [sent-361, score-0.659]
94 The simplicity, efficiency and effectiveness of this motion compensation approach make it applicable to any action recognition framework based on motion descriptors and trajectories. [sent-362, score-1.155]
95 The second contribution is the new DCS descriptor derived from the first-order scalar motion quantities specifying the local motion patterns. [sent-363, score-0.744]
96 Finally, we show that VLAD encoding technique instead of bag-of-words boosts several action descriptors, and overall exhibits a significantly better performance when combining different types of descriptors. [sent-365, score-0.38]
97 Human action recognition in videos using kinematic features and multiple instance learning. [sent-372, score-0.44]
98 Trajectory-based modeling of human actions with motion reference points. [sent-432, score-0.352]
99 Trajectons: Action recognition through the motion analysis of tracked features. [sent-489, score-0.322]
100 Action recognition in videos acquired by a moving camera using motion decomposition of lagrangian particle trajectories. [sent-612, score-0.392]
wordName wordTfidf (topN-words)
[('vlad', 0.347), ('flow', 0.271), ('motion', 0.267), ('action', 0.25), ('dcs', 0.246), ('compensated', 0.21), ('descriptors', 0.199), ('mbh', 0.184), ('curl', 0.179), ('trajectories', 0.176), ('hof', 0.17), ('kinematic', 0.166), ('shear', 0.165), ('pt', 0.154), ('trajectory', 0.154), ('compensation', 0.148), ('descriptor', 0.143), ('dominant', 0.136), ('affine', 0.095), ('encoding', 0.094), ('olympic', 0.092), ('optical', 0.09), ('waff', 0.09), ('bof', 0.086), ('actions', 0.085), ('camera', 0.071), ('divergence', 0.067), ('impact', 0.066), ('bouthemy', 0.06), ('compensating', 0.059), ('tracking', 0.058), ('yt', 0.057), ('video', 0.055), ('sake', 0.053), ('actor', 0.052), ('hollywood', 0.052), ('xt', 0.05), ('hog', 0.048), ('marzalek', 0.045), ('vorticity', 0.045), ('complementary', 0.041), ('xic', 0.04), ('xjc', 0.04), ('scalar', 0.038), ('movies', 0.038), ('vig', 0.037), ('sports', 0.037), ('laptev', 0.036), ('technique', 0.036), ('canceling', 0.035), ('gaidon', 0.035), ('clips', 0.034), ('computation', 0.034), ('residual', 0.034), ('dense', 0.033), ('codebook', 0.033), ('subfigure', 0.033), ('namely', 0.032), ('cancel', 0.032), ('hyperbolic', 0.032), ('conveys', 0.032), ('interchange', 0.032), ('jiang', 0.032), ('tracked', 0.031), ('computed', 0.031), ('moving', 0.03), ('nx', 0.03), ('div', 0.03), ('quantities', 0.029), ('involved', 0.028), ('art', 0.028), ('clues', 0.027), ('ny', 0.027), ('state', 0.026), ('aser', 0.026), ('ullah', 0.026), ('encode', 0.026), ('movement', 0.026), ('enhancing', 0.026), ('volume', 0.025), ('motions', 0.025), ('improves', 0.025), ('actors', 0.025), ('convey', 0.025), ('horizontal', 0.025), ('combination', 0.024), ('vertical', 0.024), ('reports', 0.024), ('recognition', 0.024), ('flows', 0.024), ('components', 0.024), ('improvement', 0.024), ('multi', 0.023), ('svm', 0.023), ('physical', 0.023), ('stages', 0.023), ('multiresolution', 0.023), ('quantized', 0.022), ('successive', 0.022), ('kinds', 0.022)]
simIndex simValue paperId paperTitle
same-paper 1 1.0000002 59 cvpr-2013-Better Exploiting Motion for Better Action Recognition
Author: Mihir Jain, Hervé Jégou, Patrick Bouthemy
Abstract: Several recent works on action recognition have attested the importance of explicitly integrating motion characteristics in the video description. This paper establishes that adequately decomposing visual motion into dominant and residual motions, both in the extraction of the space-time trajectories and for the computation of descriptors, significantly improves action recognition algorithms. Then, we design a new motion descriptor, the DCS descriptor, based on differential motion scalar quantities, divergence, curl and shear features. It captures additional information on the local motion patterns enhancing results. Finally, applying the recent VLAD coding technique proposed in image retrieval provides a substantial improvement for action recognition. Our three contributions are complementary and lead to outperform all reported results by a significant margin on three challenging datasets, namely Hollywood 2, HMDB51 and Olympic Sports. 1. Introduction and related work Human actions often convey the essential meaningful content in videos. Yet, recognizing human actions in un- constrained videos is a challenging problem in Computer Vision which receives a sustained attention due to the potential applications. In particular, there is a large interest in designing video-surveillance systems, providing some automatic annotation of video archives as well as improving human-computer interaction. The solutions proposed to address this problem inherit, to a large extent, from the techniques first designed for the goal of image search and classification. The successful local features developed to describe image patches [15, 23] have been translated in the 2D+t domain as spatio-temporal local descriptors [13, 30] and now include motion clues [29]. These descriptors are often extracted from spatial-temporal interest points [12, 3 1]. More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32]. Most of these approaches produce large set of local descriptors which are in turn aggregated to produce a single vector representing the video, in order to enable the use of powerful discriminative classifiers such as support vector machines (SVMs). This is usually done with the bag- Figure 1. Optical flow field vectors (green vectors with red end points) before and after dominant motion compensation. Most of the flow vectors due to camera motion are suppressed after compensation. One of the contributions of this paper is to show that compensating for the dominant motion is beneficial for most of the existing descriptors used for action recognition. of-words technique [24], which quantizes the local features using a k-means codebook. Thanks to the successful combination of this encoding technique with the aforementioned local descriptors, the state of the art in action recognition is able to go beyond the toy problems ofclassifying simple human actions in controlled environment and considers the detection of actions in real movies or video clips [11, 16]. Despite these progresses, the existing descriptors suffer from an uncompleted handling of motion in the video sequence. Motion is arguably the most reliable source of information for action recognition, as often related to the actions of interest. However, it inevitably involves the background or camera motion when dealing with uncontrolled and re- alistic situations. Although some attempts have been made to compensate camera motion in several ways [10, 21, 26, 29, 32], how to separate action motion from that caused by the camera, and how to reflect it in the video description remains an open issue. The motion compensation mechanism employed in [10] is tailor-made to the Motion Interchange Pattern encoding technique. The Motion Boundary Histogram (MBH) [29] is a recent appealing approach to 222555555533 suppress the constant motion by considering the flow gradient. It is robust to some extent to the presence of camera motion, yet it does not explicitly handle the camera motion. Another approach [26] uses a sophisticated and robust (RANSAC) estimation of camera motion. It first segments the color image into regions corresponding to planar parts in the scene and estimates the (three) dominant homographies to update the motion associated with local features. A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level. All these works support the potential of motion compensation. As the first contribution of this paper, we address the problem in a way that departs from these works by considering the compensation of the dominant motion in both the tracking stages and encoding stages involved in the computation of action recognition descriptors. We rely on the pioneering works on motion compensation such as the technique proposed in [20], that considers 2D polynomial affine motion models for estimating the dominant image motion. We consider this particular model for its robustness and its low computational cost. It was already used in [21] to separate the dominant motion (assumed to be due to the camera motion) and the residual motion (corresponding to the independent scene motions) for dynamic event recognition in videos. However, the statistical modeling of both motion components was global (over the entire image) and only the normal flow was computed for the latter. Figure 1 shows the vectors of optical flow before and after applying the proposed motion compensation. Our method successfully suppresses most of the background motion and reinforces the focus towards the action of interest. We exploit this compensated motion both for descriptor computation and for extracting trajectories. However, we also show that the camera motion should not be thrown as it contains complementary information that is worth using to recognize certain action categories. Then, we introduce the Divergence-Curl-Shear (DCS) descriptor, which encodes scalar first-order motion features, namely the motion divergence, curl and shear. It captures physical properties of the flow pattern that are not involved in the best existing descriptors for action recognition, except in the work of [1] which exploits divergence and vorticity among a set of eleven kinematic features computed from the optical flow. Our DCS descriptor provides a good performance recognition performance on its own. Most importantly, it conveys some information which is not captured by existing descriptors and further improves the recognition performance when combined with the other descriptors. As a last contribution, we bring an encoding technique known as VLAD (vector oflocal aggregated descriptors) [8] to the field of action recognition. This technique is shown to be better than the bag-of-words representation for combining all the local video descriptors we have considered. The organization of the paper is as follows. Section 2 introduces the motion properties that we will consider through this paper. Section 3 presents the datasets and classification scheme used in our different evaluations. Section 4 details how we revisit several popular descriptors of the literature by the means of dominant motion compensation. Our DCS descriptor based on kinematic properties is introduced in Section 5 and improved by the VLAD encoding technique, which is introduced and bench-marked in Section 6 for several video descriptors. Section 7 provides a comparison showing the large improvement achieved over the state of the art. Finally, Section 8 concludes the paper. 2. Motion Separation and Kinematic Features In this section, we describe the motion clues we incorporate in our action recognition framework. We separate the dominant motion and the residual motion. In most cases, this will account to distinguishing the impact of camera movement and independent actions. Note that we do not aim at recovering the 3D camera motion: The 2D parametric motion model describes the global (or dominant) motion between successive frames. We first explain how we estimate the dominant motion and employ it to separate the dominant flow from the optical flow. Then, we will introduce kinematic features, namely divergence, curl and shear for a more comprehensive description of the visual motion. 2.1. Affine motion for compensating camera motion Among polynomial motion models, we consider the 2D affine motion model. Simplest motion models such as the 4parameter model formed by the combination of 2D translation, 2D rotation and scaling, or more complex ones such as the 8-parameter quadratic model (equivalent to a homography), could be selected as well. The affine model is a good trade-off between accuracy and efficiency which is of primary importance when processing a huge video database. It does have limitations since strictly speaking it implies a single plane assumption for the static background. However, this is not that penalizing (especially for outdoor scenes) if differences in depth remain moderated with respect to the distance to the camera. The affine flow vector at point p = (x, y) and at time t, is defined as waff(pt) =?cc12((t ) ?+?aa31((t ) aa42((t ) ? ?xytt?. (1) = + + = + uaff(pt) c1(t) a1(t)xt a2(t)yt and vaff(pt) c2(t) a3 (t)xt + a4(t)yt are horizontal and vertical components of waff(pt) respectively. Let us denote the optical flow vector at point p at time t as w(pt) = (u(pt) , v(pt)). We introduce the flow vector ω(pt) obtained by removing the affine flow vector from the optical flow vector ω(pt) = w(pt) − waff(pt) . (2) 222555555644 The dominant motion (estimated as waff(pt)) is usually due to the camera motion. In this case, Equation 2 amounts to canceling (or compensating) the camera motion. Note that this is not always true. For example in case of close-up on a moving actor, the dominant motion will be the affine estimation of the apparent actor motion. The interpretation of the motion compensation output will not be that straightforward in this case, however the resulting ω-field will still exhibit different patterns for the foreground action part and the background part. In the remainder, we will refer to the “compensated” flow as ω-flow. Figure 1 displays the computed optical flow and the ωflow. We compute the affine flow with the publicly available Motion2D software1 [20] which implements a realtime robust multiresolution incremental estimation framework. The affine motion model has correctly accounted for the motion induced by the camera movement which corresponds to the dominant motion in the image pair. Indeed, we observe that the compensated flow vectors in the background are close to null and the compensated flow in the foreground, i.e., corresponding to the actors, is conversely inflated. The experiments presented along this paper will show that effective separation of dominant motion from the residual motions is beneficial for action recognition. As explained in Section 4, we will compute local motion descriptors, such as HOF, on both the optical flow and the compensated flow (ω-flow), which allows us to explicitly and directly characterize the scene motion. 2.2. Local kinematic features By kinematic features, we mean local first-order differential scalar quantities computed on the flow field. We consider the divergence, the curl (or vorticity) and the hyperbolic terms. They inform on the physical pattern of the flow so that they convey useful information on actions in videos. They can be computed from the first-order derivatives of the flow at every point p at every frame t as ⎨⎪ ⎪ ⎪ ⎧hcdyuipvr1l2(p t) = −∂ u ∂ (yxp(xtp) +−∂ v ∂ v(px ypxt ) The diverg⎪⎩ence is related to axial motion, expansion scaling effects, the curl to rotation in the image plane. hyperbolic terms express the shear of the visual flow responding to more complex configuration. We take account the shear quantity only: shear(pt) = ?hyp12(pt) + hyp22(pt). (3) and The corinto (4) 1http://www.irisa.fr/vista/Motion2D/ In Section 5, we propose the DCS descriptor that is based on the kinematic features (divergence, curl and shear) of the visual motion discussed in this subsection. It is computed on either the optical or the compensated flow, ω-flow. 3. Datasets and evaluation This section first introduces the datasets used for the evaluation. Then, we briefly present the bag-of-feature model and the classification scheme used to encode the descriptors which will be introduced in Section 4. Hollywood2. The Hollywood2 dataset [16] contains 1,707 video clips from 69 movies representing 12 action classes. It is divided into train set and test set of 823 and 884 samples respectively. Following the standard evaluation protocol of this benchmark, we use average precision (AP) for each class and the mean of APs (mAP) for evaluation. HMDB51. The HMDB51 dataset [11] is a large dataset containing 6,766 video clips extracted from various sources, ranging from movies to YouTube. It consists of 51 action classes, each having at least 101 samples. We follow the evaluation protocol of [11] and use three train/test splits, each with 70 training and 30 testing samples per class. The average classification accuracy is computed over all classes. Out of the two released sets, we use the original set as it is more challenging and used by most of the works reporting results in action recognition. Olympic Sports. The third dataset we use is Olympic Sports [19], which again is obtained from YouTube. This dataset contains 783 samples with 16 sports action classes. We use the provided2 train/test split, there are 17 to 56 training samples and 4 to 11test samples per class. Mean AP is used for the evaluation, which is the standard choice. Bag of features and classification setup. We first adopt the standard BOF [24] approach to encode all kinds of descriptors. It produces a vector that serves as the video representation. The codebook is constructed for each type of descriptor separately by the k-means algorithm. Following a common practice in the literature [27, 29, 30], the codebook size is set to k=4,000 elements. Note that Section 6 will consider encoding technique for descriptors. For the classification, we use a non-linear SVM with χ2kernel. When combining different descriptors, we simply add the kernel matrices, as done in [27]: K(xi,xj) = exp?−?cγ1cD(xic,xjc)?, 2http://vision.stanford.edu/Datasets/OlympicSports/ 222555555755 (5) where D(xic, xjc) is χ2 distance between video xic and xjc with respect to c-th channel, corresponding to c-th descriptor. The quantity γc is the mean value of χ2 distances between the training samples for the c-th channel. The multiclass classification problem that we consider is addressed by applying a one-against-rest approach. 4. Compensated descriptors This section describes how the compensation ofthe dominant motion is exploited to improve the quality of descriptors encoding the motion and the appearance around spatio-temporal positions, hence the term “compensated descriptors”. First, we briefly review the local descriptors [5, 13, 16, 29, 30] used here along with dense trajectories [29]. Second, we analyze the impact of motion flow compensation when used in two different stages of the descriptor computation, namely in the tracking and the description part. 4.1. Dense trajectories and local descriptors Employing dense trajectories to compute local descriptors is one of the state-of-the-art approaches for action recognition. It has been shown [29] that when local descriptors are computed over dense trajectories the performance improves considerably compared to when computed over spatio temporal features [30]. Dense Trajectories [29]: The trajectories are obtained by densely tracking sampled points using optical flow fields. First, feature points are sampled from a dense grid, with step size of 5 pixels and over 8 scales. Each feature point pt = (xt, yt) at frame t is then tracked to the next frame by median filtering in a dense optical flow field F = (ut, vt) as follows: pt+1 = (xt+1 , yt+1) = (xt, yt) + (M ∗ F) | (x ¯t,y ¯t) , (6) where M is the kernel of median filtering and ( x¯ t, y¯ t) is the rounded position of (xt, yt). The tracking is limited to L (=15) frames to avoid any drifting effect. Excessively short trajectories and trajectories exhibiting sudden large displacements are removed as they induce some artifacts. Trajectories must be understood here as tracks in the spacetime volume of the video. Local descriptors: The descriptors are computed within a space-time volume centered around each trajectory. Four types of descriptors are computed to encode the shape of the trajectory, local motion pattern and appearance, namely Trajectory [29], HOF (histograms of optical flow) [13], MBH [4] and HOG (histograms of oriented gradients) [3]. All these descriptors depend on the flow field used for the tracking and as input of the descriptor computation: 1. The Trajectory descriptor encodes the shape of the trajectory represented by the normalized relative coor- × dinates of the successive points forming the trajectory. It directly depends on the dense flow used for tracking points. 2. HOF is computed using the orientations and magnitudes of the flow field. 3. MBH is designed to capture the gradient of horizontal and vertical components of the flow. The motion boundaries encode the relative pixel motion and therefore suppress camera motion, but only to some extent. 4. HOG encodes the appearance by using the intensity gradient orientations and magnitudes. It is formally not a motion descriptor. Yet the position where the descriptor is computed depends on the trajectory shape. As in [29], volume around a feature point is divided into a 2 2 3 space-time grid. The orientations are quantized ian 2to × ×8 b2i ×ns 3fo srp HacOe-Gti amned g g9r ibdi.ns T fhoer o oHriOenFt (awtioitnhs one a qdudainttiiozneadl zero bin). The horizontal and vertical components of MBH are separately quantized into 8 bins each. 4.2. Impact of motion compensation The optical flow is simply referred to as flow in the following, while the compensated flow (see subsection 2. 1) is denoted by ω-flow. Both of them are considered in the tracking and descriptor computation stages. The trajectories obtained by tracking with the ω-flow are called ω-trajectories. Figure 2 comparatively illustrates the ωtrajectories and the trajectories obtained using the flow. The input video shows a man moving away from the car. In this video excerpt, the camera is following the man walking to the right, thus inducing a global motion to the left in the video. When using the flow, the computed trajectories reflect the combination of these two motion components (camera and scene motion) as depicted by Subfigure 2(b), which hampers the characterization of the current action. In contrast, the ω-trajectories plotted in Subfigure 2(c) are more active on the actor moving on the foreground, while those localized in the background are now parallel to the time axis enhancing static parts of the scene. The ω-trajectories are therefore more relevant for action recognition, since they are more regularly and more exclusively following the actor’s motion. Impact on Trajectory and HOG descriptors. Table 1reports the impact of ω-trajectories on Trajectory and HOG descriptors, which are both significantly improved by 3%4% of mAP on the two datasets. When improved by ωflow, these descriptors will be respectively referred to as ω-Trajdesc and ω-HOG in the rest of the paper. Although the better performance of ω-Trajdesc versus the original Trajectory descriptor was expected, the one 222555555866 2. Trajectories obtained from optical and compensated flows. The green tail is the trajectory the current frame. The trajectories are sub-sampled for the sake of clarity. The frames are extracted Figure over every 15 frames with red dot indicating 5 frames in this example. DescriptorHollywood2HMDB51 BaseTrliaωnje- Tc(rtoarejrdpyreos[c2d9u]ced)54 7 1. 7 4% %2382.–89% BaseliHnωOe- (GHreOp [2rG9od]uced)4 451 . 658%%%2296.– 13%% Table 1. ω-Trajdesc and ω-HOG: Impact of compensating flow on Trajectory descriptor and HOG descriptors. achieved by ω-HOG might be surprising. Our interpretation is that HOG captures more context with the modified trajectories. More precisely, the original HOG descriptor is computed from a 2D+t sub-volume aligned with the corresponding trajectory and hence represents the appearance along the trajectory shape. When using ω-flow, we do not align the video sequence. As a result, the ω-HOG descriptor is no more computed around the very same tracked physical point in the space-time volume but around points lying in a patch of the initial feature point, whose size depends on the affine flow magnitude. ω-HOG can be viewed as a “patchbased” computation capturing more information about the appearance of the background or of the moving foreground. As for ω-trajectories, they are closer to the real trajectories of the moving actors as they usually cancel the camera movement, and so, more easier to train and recognize. Impact on HOF. The ω-flow impacts computation used as an input to HOF computation itself. Therefore, HOF can both types of trajectories (ω-trajectories both the trajectory and the descriptor be computed along or those extracted MethodHollywood2HMDB51 Table(ω2rHf.-alocO IwomkF)inpHgacOtFobf[2u9ωhsb]i:f-nlo ωgwot-hwωHOflFown5H 02 34O. 58291F% %descripto3 r706s38.:–1076% m%APfor Hollywood2 and average accuracy for HMDB5 1. The ω-HOF is used in subsequent evaluations. from flow) and can encode both kinds of flows (ω-flow or flow). For the sake of completeness, we evaluate all the variants as well as the combination of both flows in the descriptor computation stage. The results are presented in Table 2 and demonstrate the significant improvement obtained by computing the HOF descriptor with the ω-flow instead of the optical flow. Note that the type of trajectories which is used, either “Tracking flow” or “Tracking ω-flow”, has a limited impact in this case. From now on, we only consider the “Tracking ω-flow” case where HOF is computed along ω-trajectories. Interestingly, combining the HOF computed from the flow and the ω-flow further improves the results. This suggests that the two flow fields are complementary and the affine flow that was subtracted from ω-flow brings in additional information. For the sake of brevity, the combination of the two kinds of HOF, i.e., computed from the flow and the ω-flow using ω-trajectories, is referred to as the ω-HOF 222555555977 MethodHollywood2HMDB51 Tab(lerT3a.cIkmM inpBgMacHgωtBf-loH w [u2)s9in]gω f-lo wo MBH5 d42 e.052s7c% riptos:m34A90P.–3769f% orHllywood2 and average accuracy for HMDB5 1. DTerHasMjcBrOeblitpHGeFor4.ySumTωraw- frcilykto hw ionfgtheduωpCs-fcaolrtωmeiwNp-df/tl+Aωoutrw-finlogwthdesωcr- isTpc-fHtrMloaiOjrBpdswtGeHFosrc descriptor in the rest of this paper. Compared to the HOF baseline, the ω-HOF descriptor achieves a gain of +3.1% of mAP on Hollywood 2 and of +7.8% on HMDB51. Impact on MBH. Since MBH is computed from gradient of flow and cancel the constant motion, there is practically no benefit in using the ω-flow to compute the MBH descriptors, as shown in Table 3. However, by tracking ω-flow, the performance improves by around 1.3% for HMDB5 1 dataset and drops by around 1.5% for Hollywood2. This relative performance depends on the encoding technique. We will come back on this descriptor when considering another encoding scheme for local descriptors in Section 6. 4.3. Summary of compensated descriptors Table 4 summarizes the refined versions of the descriptors obtained by exploiting the ω-flow, and both ω-flow and the optical flow in the case of HOF. The revisited descriptors considerably improve the results compared to the orig- inal ones, with the noticeable exception of ω-MBH which gives mixed performance with a bag-of-features encoding scheme. But we already mention as this point that this incongruous behavior of ω-MBH is stabilized with the VLAD encoding scheme considered in Section 6. Another advantage of tracking the compensated flow is that fewer trajectories are produced. For instance, the total number of trajectories decreases by about 9. 16% and 22.81% on the Hollywood2 and HMDB51 datasets, respectively. Note that exploiting both the flow and the ω-flow do not induce much computational overhead, as the latter is obtained from the flow and the affine flow which is computed in real-time and already used to get the ω-trajectories. The only additional computational cost that we introduce by using the descriptors summarized in Table 4 is the computation of a second HOF descriptor, but this stage is relatively efficient and not the bottleneck of the extraction procedure. 5. Divergence-Curl-Shear descriptor This section introduces a new descriptor encoding the kinematic properties of motion discussed in Section 2.2. It is denoted by DCS in the rest of this paper. Combining kinematic features. The spatial derivatives are computed for the horizontal and vertical components of the flow field, which are used in turn to compute the divergence, curl and shear scalar values, see Equation 3. We consider all possible pairs of kinematic features, namely (div, curl), (div, shear) and (curl, shear). At each × ×× pixel, we compute the orientation and magnitude of the 2-D vector corresponding to each of these pairs. The orientation is quantized into histograms and the magnitude is used for weighting, similar to SIFT. Our motivation for encoding pairs is that the joint distribution of kinematic features conveys more information than exploiting them independently. Implementation details. The descriptor computation and parameters are similar to HOG and other popular descriptors such as MBH, HOF. We obtain 8-bin histograms for each of the three feature pairs or components of DCS. The range of possible angles is 2π for the (div,curl) pair and π for the other pairs, because the shear is always positive. The DCS descriptor is computed for a space-time volume aligned with a trajectory, as done with the four descriptors mentioned in the previous section. In order to capture the spatio-temporal structure of kinematic features, the volume (32 32 pixels and L = 15 frames) is subdivided into a spatio-temporal grid nofd s Lize = nx 5× f ny m×e nt, sw situhb nx =de ny =to 2a and nt = 3. These parameters ×hnave× × bneen fixed for the sake of consistency with the other descriptors. For each pair of kinematic features, each cell in the grid is represented by a histogram. The resulting local descriptors have a dimensionality equal to 288 = nx ny nt 8 3. At the video level, these descriptors are nenc×od end i×nto 8 a single vector representation using either BOF or the VLAD encoding scheme introduced in the next section. 6. VLAD in actions VLAD [8] is a descriptor encoding technique that aggregates the descriptors based on a locality criterion in the feature space. To our knowledge, this technique has never been considered for action recognition. Below, we briefly introduce this approach and give the performance achieved for all the descriptors introduced along the previous sections. VLAD in brief. Similar to BOF, VLAD relies on a codebook C = {c1, c2 , ...ck} of k centroids learned by k-means. bTohoek representation is ob}t oaifn ked c by summing, efodr b yea kch-m mveiasunasl. word ci, the differences x − ci of the vectors x assigned to ci, thereby producing a sv exct −or c representation oflength d×k, 222555556088 DMeBscHriptorV5 LH.A1o%Dlywo5Bo4d.O2 %F4V3L.3HA%MD B35B91.O7%F Taωbl-eDHM5rOCBa.FSjGPdHe+rsωfco-mMHaBOnFeofV54L2936A.51D% with5431ω208-.5T96% rajde3s42c97158,.ω3% -HOG342,58019ω.6-% HOF descriptors and their combination. where d is the dimension ofthe local descriptors. We use the codebook size, k = 256. Despite this large dimensionality, VLAD is efficient because it is effectively compared with a linear kernel. VLAD is post-processed using a componentwise power normalization, which dramatically improves its performance [8]. While cross validating the parameter α involved in this power normalization, we consistently observe, for all the descriptors, a value between 0.15 and 0.3. Therefore, this parameter is set to α = 0.2 in all our experiments. For classification, we use a linear SVM and oneagainst-rest approach everywhere, unless stated otherwise. Impact on existing descriptors. We employ VLAD because it is less sensitive to quantization parameters and appears to provide better performance with descriptors having a large dimensionality. These properties are interesting in our case, because the quantization parameters involved in the DCS and MBH descriptors have been used unchanged in Section 4 for the sake of direct comparison. They might be suboptimal when using the ω-flow instead of the optical flow on which they have initially been optimized [29]. Results for MBH and ω-MBH in Table 5 supports this argument. When using VLAD instead of BOF, the scores are stable in both the cases and there is no mixed inference as that observed in Table 3. VLAD also has significant positive influence on accuracy of ω-DCS descriptor. We also observe that ω-DCS is complementary to ω-MBH and adds to the performance. Still DCS is probably not best utilized in the current setting of parameters. In case of ω-Trajdesc and ω-HOG, the scores are better with BOF on both the datasets. ω-HOF with VLAD improves on HMDB5 1, but remains equivalent for Hollywood2. Although BOF leads to better scores for the descriptors considered individually, their combination with VLAD outperforms the BOF. 7. Comparison with the state of the art This section reports our results with all descriptors combined and compares our method with the state of the art. TrajectorCy+omHbOiGna+tHioOnF+ MDBCHSHol5 l98y.w76%o%od2H4M489.D02%B%51 All ω-descriptors all five compensated descriptors using combined62.5%52.1% Table 6. Combination of VLAD representation. WU*JVliaOnughreM tHaeolth. [yo2w9d87o] 256 0985. 37% SKa*duOeJinhrau tnegdMteHatlMa.h [ol1Dd.0B[91]25 24 609.8172% Table 7. Comparison with the state of the art on Hollywood2 and HMDB5 1 datasets. *Vig et al. [28] gets 61.9% by using external eye movements data. *Jiang et al. [9] used one-vs-one multi class SVM while our and other methods use one-vs-rest SVMs. With one-against-one multi class SVM we obtain 45. 1% for HMDB51. Descriptor combination. Table 6 reports the results obtained when the descriptors are combined. Since we use VLAD, our baseline is updated that is combination of Trajectory, HOG, HOF and MBH with VLAD representation. When DCS is added to the baseline there is an improvement of 0.9% and 1.2%. With combination of all five compensated descriptors we obtain 62.5% and 52.1% on the two datasets. This is a large improvement even over the updated baseline, which shows that the proposed motion compensation and the way we exploit it are significantly important for action recognition. The comparison with the state of the art is shown in Table 7. Our method outperforms all the previously reported results in the literature. In particular, on the HMDB51 dataset, the improvement over the best reported results to date is more than 11% in average accuracy. Jiang el al. [9] used a one-against-one multi-class SVM, which might have resulted in inferior scores. With a similar multi-class SVM approach, our method obtains 45. 1%, which remains significantly better than their result. All others results were reported with one-against-rest approach. On Olympic Sports dataset we obtain mAP of 83.2% with ‘All ω-descriptors combined’ and the improvement is mostly because of VLAD and ω-flow. The best reported mAPs on this dataset are Liu et al. [14] (74.4%) and Jiang et al. [9] (80.6%), which we exceed convincingly. Gaidon et al. [6] reports the best average accuracy of 82.7%. 8. Conclusions This paper first demonstrates the interest of canceling the dominant motion (predominantly camera motion) to make the visual motion truly related to actions, for both the trajectory extraction and descriptor computation stages. It pro222555556199 duces significantly better versions (called compensated descriptors) of several state-of-the-art local descriptors for action recognition. The simplicity, efficiency and effectiveness of this motion compensation approach make it applicable to any action recognition framework based on motion descriptors and trajectories. The second contribution is the new DCS descriptor derived from the first-order scalar motion quantities specifying the local motion patterns. It captures additional information which is proved complementary to the other descriptors. Finally, we show that VLAD encoding technique instead of bag-of-words boosts several action descriptors, and overall exhibits a significantly better performance when combining different types of descriptors. Our contributions are all complementary and significantly outperform the state of the art when combined, as demon- strated by our extensive experiments on the Hollywood 2, HMDB51 and Olympic Sports datasets. Acknowledgments This work was supported by the Quaero project, funded by Oseo, French agency for innovation. We acknowledge Heng Wang’s help for reproducing some of their results. References [1] S. Ali and M. Shah. Human action recognition in videos using kinematic features and multiple instance learning. IEEE T-PAMI, 32(2):288–303, Feb. 2010. [2] T. Brox and J. Malik. Object segmentation by long term analysis of point trajectories. In ECCV, Sep. 2010. [3] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In CVPR, Jun. 2005. [4] N. Dalal, B. Triggs, and C. Schmid. Human detection using oriented histograms of flow and appearance. In ECCV, May 2006. [5] P. Dollar, V. Rabaud, G. Cottrell, and S. Belongie. Behavior recognition via sparse spatio-temporal features. In VS-PETS, Oct. 2005. [6] A. Gaidon, Z. Harchaoui, and C. Schmid. Recognizing activities with cluster-trees of tracklets. In BMVC, Sep. 2012. [7] A. Hervieu, P. Bouthemy, and J.-P. Le Cadre. A statistical video content recognition method using invariant features on object trajectories. IEEE T-CSVT, 18(1 1): 1533–1543, 2008. [8] H. J ´egou, F. Perronnin, M. Douze, J. S ´anchez, P. P ´erez, and C. Schmid. Aggregating local descriptors into compact [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] codes. IEEE T-PAMI, 34(9):1704–1716, 2012. Y.-G. Jiang, Q. Dai, X. Xue, W. Liu, and C.-W. Ngo. Trajectory-based modeling of human actions with motion reference points. In ECCV, Oct. 2012. O. Kliper-Gross, Y. Gurovich, T. Hassner, and L. Wolf. Motion interchange patterns for action recognition in unconstrained videos. In ECCV, Oct. 2012. H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. Hmdb: A large video database for human motion recognition. In ICCV, Nov. 2011. I. Laptev and T. Lindeberg. Space-time interest points. In ICCV, Oct. 2003. I. Laptev, M. Marzalek, C. Schmid, and B. Rozenfeld. Learning realistic human actions from movies. In CVPR, Jun. 2008. J. Liu, B. Kuipers, and S. Savarese. Recognizing human actions by attributes. In CVPR, Jun. 2011. D. Lowe. Distinctive image features from scale-invariant keypoints. IJCV, 60(2):91–1 10, Nov. 2004. M. Marzalek, I. Laptev, and C. Schmid. Actions in context. In CVPR, Jun. 2009. P. Matikainen, M. Hebert, and R. Sukthankar. Trajectons: Action recognition through the motion analysis of tracked features. In Workshop on Video-Oriented Object and Event Classification, ICCV, Sep. 2009. R. Messing, C. J. Pal, and H. A. Kautz. Activity recognition using the velocity histories of tracked keypoints. In ICCV, Sep. 2009. J. C. Niebles, C.-W. Chen, and F.-F. Li. Modeling temporal structure of decomposable motion segments for activity classification. In ECCV, Sep. 2010. [20] J.-M. Odobez and P. Bouthemy. Robust multiresolution estimation of parametric motion models. Jal of Vis. Comm. and Image Representation, 6(4):348–365, Dec. 1995. [21] G. Piriou, P. Bouthemy, and J.-F. Yao. Recognition of dynamic video contents with global probabilistic models of visual motion. IEEE T-IP, 15(1 1):3417–3430, 2006. [22] S. Sadanand and J. J. Corso. Action bank: A high-level representation of activity in video. In CVPR, Jun. 2012. [23] C. Schmid and R. Mohr. Local grayvalue invariants for image retrieval. IEEE T-PAMI, 19(5):530–534, May 1997. [24] J. Sivic and A. Zisserman. Video Google: A text retrieval approach to object matching in videos. In ICCV, pages 1470– 1477, Oct. 2003. [25] J. Sun, X. Wu, S. Yan, L. F. Cheong, T.-S. Chua, and J. Li. Hierarchical spatio-temporal context modeling for action recognition. In CVPR, Jun. 2009. [26] H. Uemura, S. Ishikawa, and K. Mikolajczyk. Feature tracking and motion compensation for action recognition. In BMVC, Sep. 2008. [27] M. M. Ullah, S. N. Parizi, and I. Laptev. Improving bag-offeatures action recognition with non-local cues. In BMVC, Sep. 2010. [28] E. Vig, M. Dorr, and D. Cox. Saliency-based space-variant descriptor sampling for action recognition. In ECCV, Oct. 2012. [29] H. Wang, A. Kl¨ aser, C. Schmid, and C.-L. Liu. Action recognition by dense trajectories. In CVPR, Jun. 2011. [30] H. Wang, M. M. Ullah, A. Kl¨ aser, I. Laptev, and C. Schmid. Evaluation of local spatio-temporal features for action recognition. In BMVC, Sep. 2009. [3 1] G. Willems, T. Tuytelaars, and L. J. V. Gool. An efficient dense and scale-invariant spatio-temporal interest point detector. In ECCV, Oct. 2008. [32] S. Wu, O. Oreifej, and M. Shah. Action recognition in videos acquired by a moving camera using motion decomposition of lagrangian particle trajectories. In ICCV, Nov. 2011. 222555666200
2 0.33011037 38 cvpr-2013-All About VLAD
Author: unkown-author
Abstract: The objective of this paper is large scale object instance retrieval, given a query image. A starting point of such systems is feature detection and description, for example using SIFT. The focus of this paper, however, is towards very large scale retrieval where, due to storage requirements, very compact image descriptors are required and no information about the original SIFT descriptors can be accessed directly at run time. We start from VLAD, the state-of-the art compact descriptor introduced by J´ egou et al. [8] for this purpose, and make three novel contributions: first, we show that a simple change to the normalization method significantly improves retrieval performance; second, we show that vocabulary adaptation can substantially alleviate problems caused when images are added to the dataset after initial vocabulary learning. These two methods set a new stateof-the-art over all benchmarks investigated here for both mid-dimensional (20k-D to 30k-D) and small (128-D) descriptors. Our third contribution is a multiple spatial VLAD representation, MultiVLAD, that allows the retrieval and local- ization of objects that only extend over a small part of an image (again without requiring use of the original image SIFT descriptors).
3 0.29391602 244 cvpr-2013-Large Displacement Optical Flow from Nearest Neighbor Fields
Author: Zhuoyuan Chen, Hailin Jin, Zhe Lin, Scott Cohen, Ying Wu
Abstract: We present an optical flow algorithm for large displacement motions. Most existing optical flow methods use the standard coarse-to-fine framework to deal with large displacement motions which has intrinsic limitations. Instead, we formulate the motion estimation problem as a motion segmentation problem. We use approximate nearest neighbor fields to compute an initial motion field and use a robust algorithm to compute a set of similarity transformations as the motion candidates for segmentation. To account for deviations from similarity transformations, we add local deformations in the segmentation process. We also observe that small objects can be better recovered using translations as the motion candidates. We fuse the motion results obtained under similarity transformations and under translations together before a final refinement. Experimental validation shows that our method can successfully handle large displacement motions. Although we particularly focus on large displacement motions in this work, we make no sac- rifice in terms of overall performance. In particular, our method ranks at the top of the Middlebury benchmark.
4 0.25787711 378 cvpr-2013-Sampling Strategies for Real-Time Action Recognition
Author: Feng Shi, Emil Petriu, Robert Laganière
Abstract: Local spatio-temporal features and bag-of-features representations have become popular for action recognition. A recent trend is to use dense sampling for better performance. While many methods claimed to use dense feature sets, most of them are just denser than approaches based on sparse interest point detectors. In this paper, we explore sampling with high density on action recognition. We also investigate the impact of random sampling over dense grid for computational efficiency. We present a real-time action recognition system which integrates fast random sampling method with local spatio-temporal features extracted from a Local Part Model. A new method based on histogram intersection kernel is proposed to combine multiple channels of different descriptors. Our technique shows high accuracy on the simple KTH dataset, and achieves state-of-the-art on two very challenging real-world datasets, namely, 93% on KTH, 83.3% on UCF50 and 47.6% on HMDB51.
5 0.25033411 334 cvpr-2013-Pose from Flow and Flow from Pose
Author: Katerina Fragkiadaki, Han Hu, Jianbo Shi
Abstract: Human pose detectors, although successful in localising faces and torsos of people, often fail with lower arms. Motion estimation is often inaccurate under fast movements of body parts. We build a segmentation-detection algorithm that mediates the information between body parts recognition, and multi-frame motion grouping to improve both pose detection and tracking. Motion of body parts, though not accurate, is often sufficient to segment them from their backgrounds. Such segmentations are crucialfor extracting hard to detect body parts out of their interior body clutter. By matching these segments to exemplars we obtain pose labeled body segments. The pose labeled segments and corresponding articulated joints are used to improve the motion flow fields by proposing kinematically constrained affine displacements on body parts. The pose-based articulated motion model is shown to handle large limb rotations and displacements. Our algorithm can detect people under rare poses, frequently missed by pose detectors, showing the benefits of jointly reasoning about pose, segmentation and motion in videos.
6 0.23963054 203 cvpr-2013-Hierarchical Video Representation with Trajectory Binary Partition Tree
7 0.22957294 158 cvpr-2013-Exploring Weak Stabilization for Motion Feature Extraction
8 0.21845682 170 cvpr-2013-Fast Rigid Motion Segmentation via Incrementally-Complex Local Models
9 0.21834908 205 cvpr-2013-Hollywood 3D: Recognizing Actions in 3D Natural Scenes
10 0.2050669 287 cvpr-2013-Modeling Actions through State Changes
11 0.19195089 40 cvpr-2013-An Approach to Pose-Based Action Recognition
12 0.18215664 246 cvpr-2013-Learning Binary Codes for High-Dimensional Data Using Bilinear Projections
13 0.17737761 124 cvpr-2013-Determining Motion Directly from Normal Flows Upon the Use of a Spherical Eye Platform
14 0.17577308 355 cvpr-2013-Representing Videos Using Mid-level Discriminative Patches
15 0.17479014 444 cvpr-2013-Unconstrained Monocular 3D Human Pose Estimation by Action Detection and Cross-Modality Regression Forest
16 0.17423111 268 cvpr-2013-Leveraging Structure from Motion to Learn Discriminative Codebooks for Scalable Landmark Classification
17 0.16553734 408 cvpr-2013-Spatiotemporal Deformable Part Models for Action Detection
18 0.1641493 123 cvpr-2013-Detection of Manipulation Action Consequences (MAC)
19 0.1584819 362 cvpr-2013-Robust Monocular Epipolar Flow Estimation
20 0.15661195 233 cvpr-2013-Joint Sparsity-Based Representation and Analysis of Unconstrained Activities
topicId topicWeight
[(0, 0.264), (1, 0.028), (2, -0.01), (3, -0.193), (4, -0.304), (5, -0.038), (6, 0.001), (7, -0.149), (8, -0.127), (9, 0.091), (10, 0.065), (11, 0.123), (12, 0.12), (13, 0.012), (14, 0.191), (15, 0.05), (16, 0.031), (17, -0.13), (18, 0.007), (19, 0.047), (20, 0.141), (21, -0.16), (22, 0.04), (23, 0.094), (24, 0.068), (25, 0.164), (26, 0.053), (27, 0.059), (28, -0.084), (29, -0.086), (30, -0.01), (31, 0.03), (32, 0.036), (33, 0.081), (34, -0.055), (35, -0.05), (36, 0.017), (37, -0.013), (38, 0.015), (39, 0.005), (40, 0.022), (41, 0.089), (42, 0.023), (43, -0.011), (44, 0.022), (45, -0.023), (46, -0.009), (47, 0.053), (48, 0.056), (49, -0.061)]
simIndex simValue paperId paperTitle
same-paper 1 0.9562363 59 cvpr-2013-Better Exploiting Motion for Better Action Recognition
Author: Mihir Jain, Hervé Jégou, Patrick Bouthemy
Abstract: Several recent works on action recognition have attested the importance of explicitly integrating motion characteristics in the video description. This paper establishes that adequately decomposing visual motion into dominant and residual motions, both in the extraction of the space-time trajectories and for the computation of descriptors, significantly improves action recognition algorithms. Then, we design a new motion descriptor, the DCS descriptor, based on differential motion scalar quantities, divergence, curl and shear features. It captures additional information on the local motion patterns enhancing results. Finally, applying the recent VLAD coding technique proposed in image retrieval provides a substantial improvement for action recognition. Our three contributions are complementary and lead to outperform all reported results by a significant margin on three challenging datasets, namely Hollywood 2, HMDB51 and Olympic Sports. 1. Introduction and related work Human actions often convey the essential meaningful content in videos. Yet, recognizing human actions in un- constrained videos is a challenging problem in Computer Vision which receives a sustained attention due to the potential applications. In particular, there is a large interest in designing video-surveillance systems, providing some automatic annotation of video archives as well as improving human-computer interaction. The solutions proposed to address this problem inherit, to a large extent, from the techniques first designed for the goal of image search and classification. The successful local features developed to describe image patches [15, 23] have been translated in the 2D+t domain as spatio-temporal local descriptors [13, 30] and now include motion clues [29]. These descriptors are often extracted from spatial-temporal interest points [12, 3 1]. More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32]. Most of these approaches produce large set of local descriptors which are in turn aggregated to produce a single vector representing the video, in order to enable the use of powerful discriminative classifiers such as support vector machines (SVMs). This is usually done with the bag- Figure 1. Optical flow field vectors (green vectors with red end points) before and after dominant motion compensation. Most of the flow vectors due to camera motion are suppressed after compensation. One of the contributions of this paper is to show that compensating for the dominant motion is beneficial for most of the existing descriptors used for action recognition. of-words technique [24], which quantizes the local features using a k-means codebook. Thanks to the successful combination of this encoding technique with the aforementioned local descriptors, the state of the art in action recognition is able to go beyond the toy problems ofclassifying simple human actions in controlled environment and considers the detection of actions in real movies or video clips [11, 16]. Despite these progresses, the existing descriptors suffer from an uncompleted handling of motion in the video sequence. Motion is arguably the most reliable source of information for action recognition, as often related to the actions of interest. However, it inevitably involves the background or camera motion when dealing with uncontrolled and re- alistic situations. Although some attempts have been made to compensate camera motion in several ways [10, 21, 26, 29, 32], how to separate action motion from that caused by the camera, and how to reflect it in the video description remains an open issue. The motion compensation mechanism employed in [10] is tailor-made to the Motion Interchange Pattern encoding technique. The Motion Boundary Histogram (MBH) [29] is a recent appealing approach to 222555555533 suppress the constant motion by considering the flow gradient. It is robust to some extent to the presence of camera motion, yet it does not explicitly handle the camera motion. Another approach [26] uses a sophisticated and robust (RANSAC) estimation of camera motion. It first segments the color image into regions corresponding to planar parts in the scene and estimates the (three) dominant homographies to update the motion associated with local features. A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level. All these works support the potential of motion compensation. As the first contribution of this paper, we address the problem in a way that departs from these works by considering the compensation of the dominant motion in both the tracking stages and encoding stages involved in the computation of action recognition descriptors. We rely on the pioneering works on motion compensation such as the technique proposed in [20], that considers 2D polynomial affine motion models for estimating the dominant image motion. We consider this particular model for its robustness and its low computational cost. It was already used in [21] to separate the dominant motion (assumed to be due to the camera motion) and the residual motion (corresponding to the independent scene motions) for dynamic event recognition in videos. However, the statistical modeling of both motion components was global (over the entire image) and only the normal flow was computed for the latter. Figure 1 shows the vectors of optical flow before and after applying the proposed motion compensation. Our method successfully suppresses most of the background motion and reinforces the focus towards the action of interest. We exploit this compensated motion both for descriptor computation and for extracting trajectories. However, we also show that the camera motion should not be thrown as it contains complementary information that is worth using to recognize certain action categories. Then, we introduce the Divergence-Curl-Shear (DCS) descriptor, which encodes scalar first-order motion features, namely the motion divergence, curl and shear. It captures physical properties of the flow pattern that are not involved in the best existing descriptors for action recognition, except in the work of [1] which exploits divergence and vorticity among a set of eleven kinematic features computed from the optical flow. Our DCS descriptor provides a good performance recognition performance on its own. Most importantly, it conveys some information which is not captured by existing descriptors and further improves the recognition performance when combined with the other descriptors. As a last contribution, we bring an encoding technique known as VLAD (vector oflocal aggregated descriptors) [8] to the field of action recognition. This technique is shown to be better than the bag-of-words representation for combining all the local video descriptors we have considered. The organization of the paper is as follows. Section 2 introduces the motion properties that we will consider through this paper. Section 3 presents the datasets and classification scheme used in our different evaluations. Section 4 details how we revisit several popular descriptors of the literature by the means of dominant motion compensation. Our DCS descriptor based on kinematic properties is introduced in Section 5 and improved by the VLAD encoding technique, which is introduced and bench-marked in Section 6 for several video descriptors. Section 7 provides a comparison showing the large improvement achieved over the state of the art. Finally, Section 8 concludes the paper. 2. Motion Separation and Kinematic Features In this section, we describe the motion clues we incorporate in our action recognition framework. We separate the dominant motion and the residual motion. In most cases, this will account to distinguishing the impact of camera movement and independent actions. Note that we do not aim at recovering the 3D camera motion: The 2D parametric motion model describes the global (or dominant) motion between successive frames. We first explain how we estimate the dominant motion and employ it to separate the dominant flow from the optical flow. Then, we will introduce kinematic features, namely divergence, curl and shear for a more comprehensive description of the visual motion. 2.1. Affine motion for compensating camera motion Among polynomial motion models, we consider the 2D affine motion model. Simplest motion models such as the 4parameter model formed by the combination of 2D translation, 2D rotation and scaling, or more complex ones such as the 8-parameter quadratic model (equivalent to a homography), could be selected as well. The affine model is a good trade-off between accuracy and efficiency which is of primary importance when processing a huge video database. It does have limitations since strictly speaking it implies a single plane assumption for the static background. However, this is not that penalizing (especially for outdoor scenes) if differences in depth remain moderated with respect to the distance to the camera. The affine flow vector at point p = (x, y) and at time t, is defined as waff(pt) =?cc12((t ) ?+?aa31((t ) aa42((t ) ? ?xytt?. (1) = + + = + uaff(pt) c1(t) a1(t)xt a2(t)yt and vaff(pt) c2(t) a3 (t)xt + a4(t)yt are horizontal and vertical components of waff(pt) respectively. Let us denote the optical flow vector at point p at time t as w(pt) = (u(pt) , v(pt)). We introduce the flow vector ω(pt) obtained by removing the affine flow vector from the optical flow vector ω(pt) = w(pt) − waff(pt) . (2) 222555555644 The dominant motion (estimated as waff(pt)) is usually due to the camera motion. In this case, Equation 2 amounts to canceling (or compensating) the camera motion. Note that this is not always true. For example in case of close-up on a moving actor, the dominant motion will be the affine estimation of the apparent actor motion. The interpretation of the motion compensation output will not be that straightforward in this case, however the resulting ω-field will still exhibit different patterns for the foreground action part and the background part. In the remainder, we will refer to the “compensated” flow as ω-flow. Figure 1 displays the computed optical flow and the ωflow. We compute the affine flow with the publicly available Motion2D software1 [20] which implements a realtime robust multiresolution incremental estimation framework. The affine motion model has correctly accounted for the motion induced by the camera movement which corresponds to the dominant motion in the image pair. Indeed, we observe that the compensated flow vectors in the background are close to null and the compensated flow in the foreground, i.e., corresponding to the actors, is conversely inflated. The experiments presented along this paper will show that effective separation of dominant motion from the residual motions is beneficial for action recognition. As explained in Section 4, we will compute local motion descriptors, such as HOF, on both the optical flow and the compensated flow (ω-flow), which allows us to explicitly and directly characterize the scene motion. 2.2. Local kinematic features By kinematic features, we mean local first-order differential scalar quantities computed on the flow field. We consider the divergence, the curl (or vorticity) and the hyperbolic terms. They inform on the physical pattern of the flow so that they convey useful information on actions in videos. They can be computed from the first-order derivatives of the flow at every point p at every frame t as ⎨⎪ ⎪ ⎪ ⎧hcdyuipvr1l2(p t) = −∂ u ∂ (yxp(xtp) +−∂ v ∂ v(px ypxt ) The diverg⎪⎩ence is related to axial motion, expansion scaling effects, the curl to rotation in the image plane. hyperbolic terms express the shear of the visual flow responding to more complex configuration. We take account the shear quantity only: shear(pt) = ?hyp12(pt) + hyp22(pt). (3) and The corinto (4) 1http://www.irisa.fr/vista/Motion2D/ In Section 5, we propose the DCS descriptor that is based on the kinematic features (divergence, curl and shear) of the visual motion discussed in this subsection. It is computed on either the optical or the compensated flow, ω-flow. 3. Datasets and evaluation This section first introduces the datasets used for the evaluation. Then, we briefly present the bag-of-feature model and the classification scheme used to encode the descriptors which will be introduced in Section 4. Hollywood2. The Hollywood2 dataset [16] contains 1,707 video clips from 69 movies representing 12 action classes. It is divided into train set and test set of 823 and 884 samples respectively. Following the standard evaluation protocol of this benchmark, we use average precision (AP) for each class and the mean of APs (mAP) for evaluation. HMDB51. The HMDB51 dataset [11] is a large dataset containing 6,766 video clips extracted from various sources, ranging from movies to YouTube. It consists of 51 action classes, each having at least 101 samples. We follow the evaluation protocol of [11] and use three train/test splits, each with 70 training and 30 testing samples per class. The average classification accuracy is computed over all classes. Out of the two released sets, we use the original set as it is more challenging and used by most of the works reporting results in action recognition. Olympic Sports. The third dataset we use is Olympic Sports [19], which again is obtained from YouTube. This dataset contains 783 samples with 16 sports action classes. We use the provided2 train/test split, there are 17 to 56 training samples and 4 to 11test samples per class. Mean AP is used for the evaluation, which is the standard choice. Bag of features and classification setup. We first adopt the standard BOF [24] approach to encode all kinds of descriptors. It produces a vector that serves as the video representation. The codebook is constructed for each type of descriptor separately by the k-means algorithm. Following a common practice in the literature [27, 29, 30], the codebook size is set to k=4,000 elements. Note that Section 6 will consider encoding technique for descriptors. For the classification, we use a non-linear SVM with χ2kernel. When combining different descriptors, we simply add the kernel matrices, as done in [27]: K(xi,xj) = exp?−?cγ1cD(xic,xjc)?, 2http://vision.stanford.edu/Datasets/OlympicSports/ 222555555755 (5) where D(xic, xjc) is χ2 distance between video xic and xjc with respect to c-th channel, corresponding to c-th descriptor. The quantity γc is the mean value of χ2 distances between the training samples for the c-th channel. The multiclass classification problem that we consider is addressed by applying a one-against-rest approach. 4. Compensated descriptors This section describes how the compensation ofthe dominant motion is exploited to improve the quality of descriptors encoding the motion and the appearance around spatio-temporal positions, hence the term “compensated descriptors”. First, we briefly review the local descriptors [5, 13, 16, 29, 30] used here along with dense trajectories [29]. Second, we analyze the impact of motion flow compensation when used in two different stages of the descriptor computation, namely in the tracking and the description part. 4.1. Dense trajectories and local descriptors Employing dense trajectories to compute local descriptors is one of the state-of-the-art approaches for action recognition. It has been shown [29] that when local descriptors are computed over dense trajectories the performance improves considerably compared to when computed over spatio temporal features [30]. Dense Trajectories [29]: The trajectories are obtained by densely tracking sampled points using optical flow fields. First, feature points are sampled from a dense grid, with step size of 5 pixels and over 8 scales. Each feature point pt = (xt, yt) at frame t is then tracked to the next frame by median filtering in a dense optical flow field F = (ut, vt) as follows: pt+1 = (xt+1 , yt+1) = (xt, yt) + (M ∗ F) | (x ¯t,y ¯t) , (6) where M is the kernel of median filtering and ( x¯ t, y¯ t) is the rounded position of (xt, yt). The tracking is limited to L (=15) frames to avoid any drifting effect. Excessively short trajectories and trajectories exhibiting sudden large displacements are removed as they induce some artifacts. Trajectories must be understood here as tracks in the spacetime volume of the video. Local descriptors: The descriptors are computed within a space-time volume centered around each trajectory. Four types of descriptors are computed to encode the shape of the trajectory, local motion pattern and appearance, namely Trajectory [29], HOF (histograms of optical flow) [13], MBH [4] and HOG (histograms of oriented gradients) [3]. All these descriptors depend on the flow field used for the tracking and as input of the descriptor computation: 1. The Trajectory descriptor encodes the shape of the trajectory represented by the normalized relative coor- × dinates of the successive points forming the trajectory. It directly depends on the dense flow used for tracking points. 2. HOF is computed using the orientations and magnitudes of the flow field. 3. MBH is designed to capture the gradient of horizontal and vertical components of the flow. The motion boundaries encode the relative pixel motion and therefore suppress camera motion, but only to some extent. 4. HOG encodes the appearance by using the intensity gradient orientations and magnitudes. It is formally not a motion descriptor. Yet the position where the descriptor is computed depends on the trajectory shape. As in [29], volume around a feature point is divided into a 2 2 3 space-time grid. The orientations are quantized ian 2to × ×8 b2i ×ns 3fo srp HacOe-Gti amned g g9r ibdi.ns T fhoer o oHriOenFt (awtioitnhs one a qdudainttiiozneadl zero bin). The horizontal and vertical components of MBH are separately quantized into 8 bins each. 4.2. Impact of motion compensation The optical flow is simply referred to as flow in the following, while the compensated flow (see subsection 2. 1) is denoted by ω-flow. Both of them are considered in the tracking and descriptor computation stages. The trajectories obtained by tracking with the ω-flow are called ω-trajectories. Figure 2 comparatively illustrates the ωtrajectories and the trajectories obtained using the flow. The input video shows a man moving away from the car. In this video excerpt, the camera is following the man walking to the right, thus inducing a global motion to the left in the video. When using the flow, the computed trajectories reflect the combination of these two motion components (camera and scene motion) as depicted by Subfigure 2(b), which hampers the characterization of the current action. In contrast, the ω-trajectories plotted in Subfigure 2(c) are more active on the actor moving on the foreground, while those localized in the background are now parallel to the time axis enhancing static parts of the scene. The ω-trajectories are therefore more relevant for action recognition, since they are more regularly and more exclusively following the actor’s motion. Impact on Trajectory and HOG descriptors. Table 1reports the impact of ω-trajectories on Trajectory and HOG descriptors, which are both significantly improved by 3%4% of mAP on the two datasets. When improved by ωflow, these descriptors will be respectively referred to as ω-Trajdesc and ω-HOG in the rest of the paper. Although the better performance of ω-Trajdesc versus the original Trajectory descriptor was expected, the one 222555555866 2. Trajectories obtained from optical and compensated flows. The green tail is the trajectory the current frame. The trajectories are sub-sampled for the sake of clarity. The frames are extracted Figure over every 15 frames with red dot indicating 5 frames in this example. DescriptorHollywood2HMDB51 BaseTrliaωnje- Tc(rtoarejrdpyreos[c2d9u]ced)54 7 1. 7 4% %2382.–89% BaseliHnωOe- (GHreOp [2rG9od]uced)4 451 . 658%%%2296.– 13%% Table 1. ω-Trajdesc and ω-HOG: Impact of compensating flow on Trajectory descriptor and HOG descriptors. achieved by ω-HOG might be surprising. Our interpretation is that HOG captures more context with the modified trajectories. More precisely, the original HOG descriptor is computed from a 2D+t sub-volume aligned with the corresponding trajectory and hence represents the appearance along the trajectory shape. When using ω-flow, we do not align the video sequence. As a result, the ω-HOG descriptor is no more computed around the very same tracked physical point in the space-time volume but around points lying in a patch of the initial feature point, whose size depends on the affine flow magnitude. ω-HOG can be viewed as a “patchbased” computation capturing more information about the appearance of the background or of the moving foreground. As for ω-trajectories, they are closer to the real trajectories of the moving actors as they usually cancel the camera movement, and so, more easier to train and recognize. Impact on HOF. The ω-flow impacts computation used as an input to HOF computation itself. Therefore, HOF can both types of trajectories (ω-trajectories both the trajectory and the descriptor be computed along or those extracted MethodHollywood2HMDB51 Table(ω2rHf.-alocO IwomkF)inpHgacOtFobf[2u9ωhsb]i:f-nlo ωgwot-hwωHOflFown5H 02 34O. 58291F% %descripto3 r706s38.:–1076% m%APfor Hollywood2 and average accuracy for HMDB5 1. The ω-HOF is used in subsequent evaluations. from flow) and can encode both kinds of flows (ω-flow or flow). For the sake of completeness, we evaluate all the variants as well as the combination of both flows in the descriptor computation stage. The results are presented in Table 2 and demonstrate the significant improvement obtained by computing the HOF descriptor with the ω-flow instead of the optical flow. Note that the type of trajectories which is used, either “Tracking flow” or “Tracking ω-flow”, has a limited impact in this case. From now on, we only consider the “Tracking ω-flow” case where HOF is computed along ω-trajectories. Interestingly, combining the HOF computed from the flow and the ω-flow further improves the results. This suggests that the two flow fields are complementary and the affine flow that was subtracted from ω-flow brings in additional information. For the sake of brevity, the combination of the two kinds of HOF, i.e., computed from the flow and the ω-flow using ω-trajectories, is referred to as the ω-HOF 222555555977 MethodHollywood2HMDB51 Tab(lerT3a.cIkmM inpBgMacHgωtBf-loH w [u2)s9in]gω f-lo wo MBH5 d42 e.052s7c% riptos:m34A90P.–3769f% orHllywood2 and average accuracy for HMDB5 1. DTerHasMjcBrOeblitpHGeFor4.ySumTωraw- frcilykto hw ionfgtheduωpCs-fcaolrtωmeiwNp-df/tl+Aωoutrw-finlogwthdesωcr- isTpc-fHtrMloaiOjrBpdswtGeHFosrc descriptor in the rest of this paper. Compared to the HOF baseline, the ω-HOF descriptor achieves a gain of +3.1% of mAP on Hollywood 2 and of +7.8% on HMDB51. Impact on MBH. Since MBH is computed from gradient of flow and cancel the constant motion, there is practically no benefit in using the ω-flow to compute the MBH descriptors, as shown in Table 3. However, by tracking ω-flow, the performance improves by around 1.3% for HMDB5 1 dataset and drops by around 1.5% for Hollywood2. This relative performance depends on the encoding technique. We will come back on this descriptor when considering another encoding scheme for local descriptors in Section 6. 4.3. Summary of compensated descriptors Table 4 summarizes the refined versions of the descriptors obtained by exploiting the ω-flow, and both ω-flow and the optical flow in the case of HOF. The revisited descriptors considerably improve the results compared to the orig- inal ones, with the noticeable exception of ω-MBH which gives mixed performance with a bag-of-features encoding scheme. But we already mention as this point that this incongruous behavior of ω-MBH is stabilized with the VLAD encoding scheme considered in Section 6. Another advantage of tracking the compensated flow is that fewer trajectories are produced. For instance, the total number of trajectories decreases by about 9. 16% and 22.81% on the Hollywood2 and HMDB51 datasets, respectively. Note that exploiting both the flow and the ω-flow do not induce much computational overhead, as the latter is obtained from the flow and the affine flow which is computed in real-time and already used to get the ω-trajectories. The only additional computational cost that we introduce by using the descriptors summarized in Table 4 is the computation of a second HOF descriptor, but this stage is relatively efficient and not the bottleneck of the extraction procedure. 5. Divergence-Curl-Shear descriptor This section introduces a new descriptor encoding the kinematic properties of motion discussed in Section 2.2. It is denoted by DCS in the rest of this paper. Combining kinematic features. The spatial derivatives are computed for the horizontal and vertical components of the flow field, which are used in turn to compute the divergence, curl and shear scalar values, see Equation 3. We consider all possible pairs of kinematic features, namely (div, curl), (div, shear) and (curl, shear). At each × ×× pixel, we compute the orientation and magnitude of the 2-D vector corresponding to each of these pairs. The orientation is quantized into histograms and the magnitude is used for weighting, similar to SIFT. Our motivation for encoding pairs is that the joint distribution of kinematic features conveys more information than exploiting them independently. Implementation details. The descriptor computation and parameters are similar to HOG and other popular descriptors such as MBH, HOF. We obtain 8-bin histograms for each of the three feature pairs or components of DCS. The range of possible angles is 2π for the (div,curl) pair and π for the other pairs, because the shear is always positive. The DCS descriptor is computed for a space-time volume aligned with a trajectory, as done with the four descriptors mentioned in the previous section. In order to capture the spatio-temporal structure of kinematic features, the volume (32 32 pixels and L = 15 frames) is subdivided into a spatio-temporal grid nofd s Lize = nx 5× f ny m×e nt, sw situhb nx =de ny =to 2a and nt = 3. These parameters ×hnave× × bneen fixed for the sake of consistency with the other descriptors. For each pair of kinematic features, each cell in the grid is represented by a histogram. The resulting local descriptors have a dimensionality equal to 288 = nx ny nt 8 3. At the video level, these descriptors are nenc×od end i×nto 8 a single vector representation using either BOF or the VLAD encoding scheme introduced in the next section. 6. VLAD in actions VLAD [8] is a descriptor encoding technique that aggregates the descriptors based on a locality criterion in the feature space. To our knowledge, this technique has never been considered for action recognition. Below, we briefly introduce this approach and give the performance achieved for all the descriptors introduced along the previous sections. VLAD in brief. Similar to BOF, VLAD relies on a codebook C = {c1, c2 , ...ck} of k centroids learned by k-means. bTohoek representation is ob}t oaifn ked c by summing, efodr b yea kch-m mveiasunasl. word ci, the differences x − ci of the vectors x assigned to ci, thereby producing a sv exct −or c representation oflength d×k, 222555556088 DMeBscHriptorV5 LH.A1o%Dlywo5Bo4d.O2 %F4V3L.3HA%MD B35B91.O7%F Taωbl-eDHM5rOCBa.FSjGPdHe+rsωfco-mMHaBOnFeofV54L2936A.51D% with5431ω208-.5T96% rajde3s42c97158,.ω3% -HOG342,58019ω.6-% HOF descriptors and their combination. where d is the dimension ofthe local descriptors. We use the codebook size, k = 256. Despite this large dimensionality, VLAD is efficient because it is effectively compared with a linear kernel. VLAD is post-processed using a componentwise power normalization, which dramatically improves its performance [8]. While cross validating the parameter α involved in this power normalization, we consistently observe, for all the descriptors, a value between 0.15 and 0.3. Therefore, this parameter is set to α = 0.2 in all our experiments. For classification, we use a linear SVM and oneagainst-rest approach everywhere, unless stated otherwise. Impact on existing descriptors. We employ VLAD because it is less sensitive to quantization parameters and appears to provide better performance with descriptors having a large dimensionality. These properties are interesting in our case, because the quantization parameters involved in the DCS and MBH descriptors have been used unchanged in Section 4 for the sake of direct comparison. They might be suboptimal when using the ω-flow instead of the optical flow on which they have initially been optimized [29]. Results for MBH and ω-MBH in Table 5 supports this argument. When using VLAD instead of BOF, the scores are stable in both the cases and there is no mixed inference as that observed in Table 3. VLAD also has significant positive influence on accuracy of ω-DCS descriptor. We also observe that ω-DCS is complementary to ω-MBH and adds to the performance. Still DCS is probably not best utilized in the current setting of parameters. In case of ω-Trajdesc and ω-HOG, the scores are better with BOF on both the datasets. ω-HOF with VLAD improves on HMDB5 1, but remains equivalent for Hollywood2. Although BOF leads to better scores for the descriptors considered individually, their combination with VLAD outperforms the BOF. 7. Comparison with the state of the art This section reports our results with all descriptors combined and compares our method with the state of the art. TrajectorCy+omHbOiGna+tHioOnF+ MDBCHSHol5 l98y.w76%o%od2H4M489.D02%B%51 All ω-descriptors all five compensated descriptors using combined62.5%52.1% Table 6. Combination of VLAD representation. WU*JVliaOnughreM tHaeolth. [yo2w9d87o] 256 0985. 37% SKa*duOeJinhrau tnegdMteHatlMa.h [ol1Dd.0B[91]25 24 609.8172% Table 7. Comparison with the state of the art on Hollywood2 and HMDB5 1 datasets. *Vig et al. [28] gets 61.9% by using external eye movements data. *Jiang et al. [9] used one-vs-one multi class SVM while our and other methods use one-vs-rest SVMs. With one-against-one multi class SVM we obtain 45. 1% for HMDB51. Descriptor combination. Table 6 reports the results obtained when the descriptors are combined. Since we use VLAD, our baseline is updated that is combination of Trajectory, HOG, HOF and MBH with VLAD representation. When DCS is added to the baseline there is an improvement of 0.9% and 1.2%. With combination of all five compensated descriptors we obtain 62.5% and 52.1% on the two datasets. This is a large improvement even over the updated baseline, which shows that the proposed motion compensation and the way we exploit it are significantly important for action recognition. The comparison with the state of the art is shown in Table 7. Our method outperforms all the previously reported results in the literature. In particular, on the HMDB51 dataset, the improvement over the best reported results to date is more than 11% in average accuracy. Jiang el al. [9] used a one-against-one multi-class SVM, which might have resulted in inferior scores. With a similar multi-class SVM approach, our method obtains 45. 1%, which remains significantly better than their result. All others results were reported with one-against-rest approach. On Olympic Sports dataset we obtain mAP of 83.2% with ‘All ω-descriptors combined’ and the improvement is mostly because of VLAD and ω-flow. The best reported mAPs on this dataset are Liu et al. [14] (74.4%) and Jiang et al. [9] (80.6%), which we exceed convincingly. Gaidon et al. [6] reports the best average accuracy of 82.7%. 8. Conclusions This paper first demonstrates the interest of canceling the dominant motion (predominantly camera motion) to make the visual motion truly related to actions, for both the trajectory extraction and descriptor computation stages. It pro222555556199 duces significantly better versions (called compensated descriptors) of several state-of-the-art local descriptors for action recognition. The simplicity, efficiency and effectiveness of this motion compensation approach make it applicable to any action recognition framework based on motion descriptors and trajectories. The second contribution is the new DCS descriptor derived from the first-order scalar motion quantities specifying the local motion patterns. It captures additional information which is proved complementary to the other descriptors. Finally, we show that VLAD encoding technique instead of bag-of-words boosts several action descriptors, and overall exhibits a significantly better performance when combining different types of descriptors. Our contributions are all complementary and significantly outperform the state of the art when combined, as demon- strated by our extensive experiments on the Hollywood 2, HMDB51 and Olympic Sports datasets. Acknowledgments This work was supported by the Quaero project, funded by Oseo, French agency for innovation. We acknowledge Heng Wang’s help for reproducing some of their results. References [1] S. Ali and M. Shah. Human action recognition in videos using kinematic features and multiple instance learning. IEEE T-PAMI, 32(2):288–303, Feb. 2010. [2] T. Brox and J. Malik. Object segmentation by long term analysis of point trajectories. In ECCV, Sep. 2010. [3] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In CVPR, Jun. 2005. [4] N. Dalal, B. Triggs, and C. Schmid. Human detection using oriented histograms of flow and appearance. In ECCV, May 2006. [5] P. Dollar, V. Rabaud, G. Cottrell, and S. Belongie. Behavior recognition via sparse spatio-temporal features. In VS-PETS, Oct. 2005. [6] A. Gaidon, Z. Harchaoui, and C. Schmid. Recognizing activities with cluster-trees of tracklets. In BMVC, Sep. 2012. [7] A. Hervieu, P. Bouthemy, and J.-P. Le Cadre. A statistical video content recognition method using invariant features on object trajectories. IEEE T-CSVT, 18(1 1): 1533–1543, 2008. [8] H. J ´egou, F. Perronnin, M. Douze, J. S ´anchez, P. P ´erez, and C. Schmid. Aggregating local descriptors into compact [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] codes. IEEE T-PAMI, 34(9):1704–1716, 2012. Y.-G. Jiang, Q. Dai, X. Xue, W. Liu, and C.-W. Ngo. Trajectory-based modeling of human actions with motion reference points. In ECCV, Oct. 2012. O. Kliper-Gross, Y. Gurovich, T. Hassner, and L. Wolf. Motion interchange patterns for action recognition in unconstrained videos. In ECCV, Oct. 2012. H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. Hmdb: A large video database for human motion recognition. In ICCV, Nov. 2011. I. Laptev and T. Lindeberg. Space-time interest points. In ICCV, Oct. 2003. I. Laptev, M. Marzalek, C. Schmid, and B. Rozenfeld. Learning realistic human actions from movies. In CVPR, Jun. 2008. J. Liu, B. Kuipers, and S. Savarese. Recognizing human actions by attributes. In CVPR, Jun. 2011. D. Lowe. Distinctive image features from scale-invariant keypoints. IJCV, 60(2):91–1 10, Nov. 2004. M. Marzalek, I. Laptev, and C. Schmid. Actions in context. In CVPR, Jun. 2009. P. Matikainen, M. Hebert, and R. Sukthankar. Trajectons: Action recognition through the motion analysis of tracked features. In Workshop on Video-Oriented Object and Event Classification, ICCV, Sep. 2009. R. Messing, C. J. Pal, and H. A. Kautz. Activity recognition using the velocity histories of tracked keypoints. In ICCV, Sep. 2009. J. C. Niebles, C.-W. Chen, and F.-F. Li. Modeling temporal structure of decomposable motion segments for activity classification. In ECCV, Sep. 2010. [20] J.-M. Odobez and P. Bouthemy. Robust multiresolution estimation of parametric motion models. Jal of Vis. Comm. and Image Representation, 6(4):348–365, Dec. 1995. [21] G. Piriou, P. Bouthemy, and J.-F. Yao. Recognition of dynamic video contents with global probabilistic models of visual motion. IEEE T-IP, 15(1 1):3417–3430, 2006. [22] S. Sadanand and J. J. Corso. Action bank: A high-level representation of activity in video. In CVPR, Jun. 2012. [23] C. Schmid and R. Mohr. Local grayvalue invariants for image retrieval. IEEE T-PAMI, 19(5):530–534, May 1997. [24] J. Sivic and A. Zisserman. Video Google: A text retrieval approach to object matching in videos. In ICCV, pages 1470– 1477, Oct. 2003. [25] J. Sun, X. Wu, S. Yan, L. F. Cheong, T.-S. Chua, and J. Li. Hierarchical spatio-temporal context modeling for action recognition. In CVPR, Jun. 2009. [26] H. Uemura, S. Ishikawa, and K. Mikolajczyk. Feature tracking and motion compensation for action recognition. In BMVC, Sep. 2008. [27] M. M. Ullah, S. N. Parizi, and I. Laptev. Improving bag-offeatures action recognition with non-local cues. In BMVC, Sep. 2010. [28] E. Vig, M. Dorr, and D. Cox. Saliency-based space-variant descriptor sampling for action recognition. In ECCV, Oct. 2012. [29] H. Wang, A. Kl¨ aser, C. Schmid, and C.-L. Liu. Action recognition by dense trajectories. In CVPR, Jun. 2011. [30] H. Wang, M. M. Ullah, A. Kl¨ aser, I. Laptev, and C. Schmid. Evaluation of local spatio-temporal features for action recognition. In BMVC, Sep. 2009. [3 1] G. Willems, T. Tuytelaars, and L. J. V. Gool. An efficient dense and scale-invariant spatio-temporal interest point detector. In ECCV, Oct. 2008. [32] S. Wu, O. Oreifej, and M. Shah. Action recognition in videos acquired by a moving camera using motion decomposition of lagrangian particle trajectories. In ICCV, Nov. 2011. 222555666200
2 0.69937414 203 cvpr-2013-Hierarchical Video Representation with Trajectory Binary Partition Tree
Author: Guillem Palou, Philippe Salembier
Abstract: As early stage of video processing, we introduce an iterative trajectory merging algorithm that produces a regionbased and hierarchical representation of the video sequence, called the Trajectory Binary Partition Tree (BPT). From this representation, many analysis and graph cut techniques can be used to extract partitions or objects that are useful in the context of specific applications. In order to define trajectories and to create a precise merging algorithm, color and motion cues have to be used. Both types of informations are very useful to characterize objects but present strong differences of behavior in the spatial and the temporal dimensions. On the one hand, scenes and objects are rich in their spatial color distributions, but these distributions are rather stable over time. Object motion, on the other hand, presents simple structures and low spatial variability but may change from frame to frame. The proposed algorithm takes into account this key difference and relies on different models and associated metrics to deal with color and motion information. We show that the proposed algorithm outperforms existing hierarchical video segmentation algorithms and provides more stable and precise regions.
3 0.67617077 244 cvpr-2013-Large Displacement Optical Flow from Nearest Neighbor Fields
Author: Zhuoyuan Chen, Hailin Jin, Zhe Lin, Scott Cohen, Ying Wu
Abstract: We present an optical flow algorithm for large displacement motions. Most existing optical flow methods use the standard coarse-to-fine framework to deal with large displacement motions which has intrinsic limitations. Instead, we formulate the motion estimation problem as a motion segmentation problem. We use approximate nearest neighbor fields to compute an initial motion field and use a robust algorithm to compute a set of similarity transformations as the motion candidates for segmentation. To account for deviations from similarity transformations, we add local deformations in the segmentation process. We also observe that small objects can be better recovered using translations as the motion candidates. We fuse the motion results obtained under similarity transformations and under translations together before a final refinement. Experimental validation shows that our method can successfully handle large displacement motions. Although we particularly focus on large displacement motions in this work, we make no sac- rifice in terms of overall performance. In particular, our method ranks at the top of the Middlebury benchmark.
4 0.66548115 137 cvpr-2013-Dynamic Scene Classification: Learning Motion Descriptors with Slow Features Analysis
Author: Christian Thériault, Nicolas Thome, Matthieu Cord
Abstract: In this paper, we address the challenging problem of categorizing video sequences composed of dynamic natural scenes. Contrarily to previous methods that rely on handcrafted descriptors, we propose here to represent videos using unsupervised learning of motion features. Our method encompasses three main contributions: 1) Based on the Slow Feature Analysis principle, we introduce a learned local motion descriptor which represents the principal and more stable motion components of training videos. 2) We integrate our local motion feature into a global coding/pooling architecture in order to provide an effective signature for each video sequence. 3) We report state of the art classification performances on two challenging natural scenes data sets. In particular, an outstanding improvement of 11 % in classification score is reached on a data set introduced in 2012.
5 0.65573138 291 cvpr-2013-Motionlets: Mid-level 3D Parts for Human Motion Recognition
Author: LiMin Wang, Yu Qiao, Xiaoou Tang
Abstract: This paper proposes motionlet, a mid-level and spatiotemporal part, for human motion recognition. Motionlet can be seen as a tight cluster in motion and appearance space, corresponding to the moving process of different body parts. We postulate three key properties of motionlet for action recognition: high motion saliency, multiple scale representation, and representative-discriminative ability. Towards this goal, we develop a data-driven approach to learn motionlets from training videos. First, we extract 3D regions with high motion saliency. Then we cluster these regions and preserve the centers as candidate templates for motionlet. Finally, we examine the representative and discriminative power of the candidates, and introduce a greedy method to select effective candidates. With motionlets, we present a mid-level representation for video, called motionlet activation vector. We conduct experiments on three datasets, KTH, HMDB51, and UCF50. The results show that the proposed methods significantly outperform state-of-the-art methods.
6 0.64967269 88 cvpr-2013-Compressible Motion Fields
7 0.64798927 158 cvpr-2013-Exploring Weak Stabilization for Motion Feature Extraction
8 0.62002933 149 cvpr-2013-Evaluation of Color STIPs for Human Action Recognition
9 0.60127854 124 cvpr-2013-Determining Motion Directly from Normal Flows Upon the Use of a Spherical Eye Platform
10 0.59590226 170 cvpr-2013-Fast Rigid Motion Segmentation via Incrementally-Complex Local Models
11 0.56735355 38 cvpr-2013-All About VLAD
12 0.56114709 10 cvpr-2013-A Fully-Connected Layered Model of Foreground and Background Flow
13 0.54830378 336 cvpr-2013-Poselet Key-Framing: A Model for Human Activity Recognition
14 0.54260296 112 cvpr-2013-Dense Segmentation-Aware Descriptors
15 0.5395689 378 cvpr-2013-Sampling Strategies for Real-Time Action Recognition
16 0.5370599 334 cvpr-2013-Pose from Flow and Flow from Pose
17 0.53027093 118 cvpr-2013-Detecting Pulse from Head Motions in Video
18 0.52014792 287 cvpr-2013-Modeling Actions through State Changes
19 0.5113461 205 cvpr-2013-Hollywood 3D: Recognizing Actions in 3D Natural Scenes
20 0.50991064 362 cvpr-2013-Robust Monocular Epipolar Flow Estimation
topicId topicWeight
[(10, 0.079), (16, 0.012), (26, 0.024), (33, 0.681), (48, 0.013), (67, 0.05), (69, 0.023), (87, 0.036)]
simIndex simValue paperId paperTitle
same-paper 1 0.99934423 59 cvpr-2013-Better Exploiting Motion for Better Action Recognition
Author: Mihir Jain, Hervé Jégou, Patrick Bouthemy
Abstract: Several recent works on action recognition have attested the importance of explicitly integrating motion characteristics in the video description. This paper establishes that adequately decomposing visual motion into dominant and residual motions, both in the extraction of the space-time trajectories and for the computation of descriptors, significantly improves action recognition algorithms. Then, we design a new motion descriptor, the DCS descriptor, based on differential motion scalar quantities, divergence, curl and shear features. It captures additional information on the local motion patterns enhancing results. Finally, applying the recent VLAD coding technique proposed in image retrieval provides a substantial improvement for action recognition. Our three contributions are complementary and lead to outperform all reported results by a significant margin on three challenging datasets, namely Hollywood 2, HMDB51 and Olympic Sports. 1. Introduction and related work Human actions often convey the essential meaningful content in videos. Yet, recognizing human actions in un- constrained videos is a challenging problem in Computer Vision which receives a sustained attention due to the potential applications. In particular, there is a large interest in designing video-surveillance systems, providing some automatic annotation of video archives as well as improving human-computer interaction. The solutions proposed to address this problem inherit, to a large extent, from the techniques first designed for the goal of image search and classification. The successful local features developed to describe image patches [15, 23] have been translated in the 2D+t domain as spatio-temporal local descriptors [13, 30] and now include motion clues [29]. These descriptors are often extracted from spatial-temporal interest points [12, 3 1]. More recent techniques assume some underlying temporal motion model involving trajectories [2, 6, 7, 17, 18, 25, 29, 32]. Most of these approaches produce large set of local descriptors which are in turn aggregated to produce a single vector representing the video, in order to enable the use of powerful discriminative classifiers such as support vector machines (SVMs). This is usually done with the bag- Figure 1. Optical flow field vectors (green vectors with red end points) before and after dominant motion compensation. Most of the flow vectors due to camera motion are suppressed after compensation. One of the contributions of this paper is to show that compensating for the dominant motion is beneficial for most of the existing descriptors used for action recognition. of-words technique [24], which quantizes the local features using a k-means codebook. Thanks to the successful combination of this encoding technique with the aforementioned local descriptors, the state of the art in action recognition is able to go beyond the toy problems ofclassifying simple human actions in controlled environment and considers the detection of actions in real movies or video clips [11, 16]. Despite these progresses, the existing descriptors suffer from an uncompleted handling of motion in the video sequence. Motion is arguably the most reliable source of information for action recognition, as often related to the actions of interest. However, it inevitably involves the background or camera motion when dealing with uncontrolled and re- alistic situations. Although some attempts have been made to compensate camera motion in several ways [10, 21, 26, 29, 32], how to separate action motion from that caused by the camera, and how to reflect it in the video description remains an open issue. The motion compensation mechanism employed in [10] is tailor-made to the Motion Interchange Pattern encoding technique. The Motion Boundary Histogram (MBH) [29] is a recent appealing approach to 222555555533 suppress the constant motion by considering the flow gradient. It is robust to some extent to the presence of camera motion, yet it does not explicitly handle the camera motion. Another approach [26] uses a sophisticated and robust (RANSAC) estimation of camera motion. It first segments the color image into regions corresponding to planar parts in the scene and estimates the (three) dominant homographies to update the motion associated with local features. A rather different view is adopted in [32] where the motion decomposition is performed at the trajectory level. All these works support the potential of motion compensation. As the first contribution of this paper, we address the problem in a way that departs from these works by considering the compensation of the dominant motion in both the tracking stages and encoding stages involved in the computation of action recognition descriptors. We rely on the pioneering works on motion compensation such as the technique proposed in [20], that considers 2D polynomial affine motion models for estimating the dominant image motion. We consider this particular model for its robustness and its low computational cost. It was already used in [21] to separate the dominant motion (assumed to be due to the camera motion) and the residual motion (corresponding to the independent scene motions) for dynamic event recognition in videos. However, the statistical modeling of both motion components was global (over the entire image) and only the normal flow was computed for the latter. Figure 1 shows the vectors of optical flow before and after applying the proposed motion compensation. Our method successfully suppresses most of the background motion and reinforces the focus towards the action of interest. We exploit this compensated motion both for descriptor computation and for extracting trajectories. However, we also show that the camera motion should not be thrown as it contains complementary information that is worth using to recognize certain action categories. Then, we introduce the Divergence-Curl-Shear (DCS) descriptor, which encodes scalar first-order motion features, namely the motion divergence, curl and shear. It captures physical properties of the flow pattern that are not involved in the best existing descriptors for action recognition, except in the work of [1] which exploits divergence and vorticity among a set of eleven kinematic features computed from the optical flow. Our DCS descriptor provides a good performance recognition performance on its own. Most importantly, it conveys some information which is not captured by existing descriptors and further improves the recognition performance when combined with the other descriptors. As a last contribution, we bring an encoding technique known as VLAD (vector oflocal aggregated descriptors) [8] to the field of action recognition. This technique is shown to be better than the bag-of-words representation for combining all the local video descriptors we have considered. The organization of the paper is as follows. Section 2 introduces the motion properties that we will consider through this paper. Section 3 presents the datasets and classification scheme used in our different evaluations. Section 4 details how we revisit several popular descriptors of the literature by the means of dominant motion compensation. Our DCS descriptor based on kinematic properties is introduced in Section 5 and improved by the VLAD encoding technique, which is introduced and bench-marked in Section 6 for several video descriptors. Section 7 provides a comparison showing the large improvement achieved over the state of the art. Finally, Section 8 concludes the paper. 2. Motion Separation and Kinematic Features In this section, we describe the motion clues we incorporate in our action recognition framework. We separate the dominant motion and the residual motion. In most cases, this will account to distinguishing the impact of camera movement and independent actions. Note that we do not aim at recovering the 3D camera motion: The 2D parametric motion model describes the global (or dominant) motion between successive frames. We first explain how we estimate the dominant motion and employ it to separate the dominant flow from the optical flow. Then, we will introduce kinematic features, namely divergence, curl and shear for a more comprehensive description of the visual motion. 2.1. Affine motion for compensating camera motion Among polynomial motion models, we consider the 2D affine motion model. Simplest motion models such as the 4parameter model formed by the combination of 2D translation, 2D rotation and scaling, or more complex ones such as the 8-parameter quadratic model (equivalent to a homography), could be selected as well. The affine model is a good trade-off between accuracy and efficiency which is of primary importance when processing a huge video database. It does have limitations since strictly speaking it implies a single plane assumption for the static background. However, this is not that penalizing (especially for outdoor scenes) if differences in depth remain moderated with respect to the distance to the camera. The affine flow vector at point p = (x, y) and at time t, is defined as waff(pt) =?cc12((t ) ?+?aa31((t ) aa42((t ) ? ?xytt?. (1) = + + = + uaff(pt) c1(t) a1(t)xt a2(t)yt and vaff(pt) c2(t) a3 (t)xt + a4(t)yt are horizontal and vertical components of waff(pt) respectively. Let us denote the optical flow vector at point p at time t as w(pt) = (u(pt) , v(pt)). We introduce the flow vector ω(pt) obtained by removing the affine flow vector from the optical flow vector ω(pt) = w(pt) − waff(pt) . (2) 222555555644 The dominant motion (estimated as waff(pt)) is usually due to the camera motion. In this case, Equation 2 amounts to canceling (or compensating) the camera motion. Note that this is not always true. For example in case of close-up on a moving actor, the dominant motion will be the affine estimation of the apparent actor motion. The interpretation of the motion compensation output will not be that straightforward in this case, however the resulting ω-field will still exhibit different patterns for the foreground action part and the background part. In the remainder, we will refer to the “compensated” flow as ω-flow. Figure 1 displays the computed optical flow and the ωflow. We compute the affine flow with the publicly available Motion2D software1 [20] which implements a realtime robust multiresolution incremental estimation framework. The affine motion model has correctly accounted for the motion induced by the camera movement which corresponds to the dominant motion in the image pair. Indeed, we observe that the compensated flow vectors in the background are close to null and the compensated flow in the foreground, i.e., corresponding to the actors, is conversely inflated. The experiments presented along this paper will show that effective separation of dominant motion from the residual motions is beneficial for action recognition. As explained in Section 4, we will compute local motion descriptors, such as HOF, on both the optical flow and the compensated flow (ω-flow), which allows us to explicitly and directly characterize the scene motion. 2.2. Local kinematic features By kinematic features, we mean local first-order differential scalar quantities computed on the flow field. We consider the divergence, the curl (or vorticity) and the hyperbolic terms. They inform on the physical pattern of the flow so that they convey useful information on actions in videos. They can be computed from the first-order derivatives of the flow at every point p at every frame t as ⎨⎪ ⎪ ⎪ ⎧hcdyuipvr1l2(p t) = −∂ u ∂ (yxp(xtp) +−∂ v ∂ v(px ypxt ) The diverg⎪⎩ence is related to axial motion, expansion scaling effects, the curl to rotation in the image plane. hyperbolic terms express the shear of the visual flow responding to more complex configuration. We take account the shear quantity only: shear(pt) = ?hyp12(pt) + hyp22(pt). (3) and The corinto (4) 1http://www.irisa.fr/vista/Motion2D/ In Section 5, we propose the DCS descriptor that is based on the kinematic features (divergence, curl and shear) of the visual motion discussed in this subsection. It is computed on either the optical or the compensated flow, ω-flow. 3. Datasets and evaluation This section first introduces the datasets used for the evaluation. Then, we briefly present the bag-of-feature model and the classification scheme used to encode the descriptors which will be introduced in Section 4. Hollywood2. The Hollywood2 dataset [16] contains 1,707 video clips from 69 movies representing 12 action classes. It is divided into train set and test set of 823 and 884 samples respectively. Following the standard evaluation protocol of this benchmark, we use average precision (AP) for each class and the mean of APs (mAP) for evaluation. HMDB51. The HMDB51 dataset [11] is a large dataset containing 6,766 video clips extracted from various sources, ranging from movies to YouTube. It consists of 51 action classes, each having at least 101 samples. We follow the evaluation protocol of [11] and use three train/test splits, each with 70 training and 30 testing samples per class. The average classification accuracy is computed over all classes. Out of the two released sets, we use the original set as it is more challenging and used by most of the works reporting results in action recognition. Olympic Sports. The third dataset we use is Olympic Sports [19], which again is obtained from YouTube. This dataset contains 783 samples with 16 sports action classes. We use the provided2 train/test split, there are 17 to 56 training samples and 4 to 11test samples per class. Mean AP is used for the evaluation, which is the standard choice. Bag of features and classification setup. We first adopt the standard BOF [24] approach to encode all kinds of descriptors. It produces a vector that serves as the video representation. The codebook is constructed for each type of descriptor separately by the k-means algorithm. Following a common practice in the literature [27, 29, 30], the codebook size is set to k=4,000 elements. Note that Section 6 will consider encoding technique for descriptors. For the classification, we use a non-linear SVM with χ2kernel. When combining different descriptors, we simply add the kernel matrices, as done in [27]: K(xi,xj) = exp?−?cγ1cD(xic,xjc)?, 2http://vision.stanford.edu/Datasets/OlympicSports/ 222555555755 (5) where D(xic, xjc) is χ2 distance between video xic and xjc with respect to c-th channel, corresponding to c-th descriptor. The quantity γc is the mean value of χ2 distances between the training samples for the c-th channel. The multiclass classification problem that we consider is addressed by applying a one-against-rest approach. 4. Compensated descriptors This section describes how the compensation ofthe dominant motion is exploited to improve the quality of descriptors encoding the motion and the appearance around spatio-temporal positions, hence the term “compensated descriptors”. First, we briefly review the local descriptors [5, 13, 16, 29, 30] used here along with dense trajectories [29]. Second, we analyze the impact of motion flow compensation when used in two different stages of the descriptor computation, namely in the tracking and the description part. 4.1. Dense trajectories and local descriptors Employing dense trajectories to compute local descriptors is one of the state-of-the-art approaches for action recognition. It has been shown [29] that when local descriptors are computed over dense trajectories the performance improves considerably compared to when computed over spatio temporal features [30]. Dense Trajectories [29]: The trajectories are obtained by densely tracking sampled points using optical flow fields. First, feature points are sampled from a dense grid, with step size of 5 pixels and over 8 scales. Each feature point pt = (xt, yt) at frame t is then tracked to the next frame by median filtering in a dense optical flow field F = (ut, vt) as follows: pt+1 = (xt+1 , yt+1) = (xt, yt) + (M ∗ F) | (x ¯t,y ¯t) , (6) where M is the kernel of median filtering and ( x¯ t, y¯ t) is the rounded position of (xt, yt). The tracking is limited to L (=15) frames to avoid any drifting effect. Excessively short trajectories and trajectories exhibiting sudden large displacements are removed as they induce some artifacts. Trajectories must be understood here as tracks in the spacetime volume of the video. Local descriptors: The descriptors are computed within a space-time volume centered around each trajectory. Four types of descriptors are computed to encode the shape of the trajectory, local motion pattern and appearance, namely Trajectory [29], HOF (histograms of optical flow) [13], MBH [4] and HOG (histograms of oriented gradients) [3]. All these descriptors depend on the flow field used for the tracking and as input of the descriptor computation: 1. The Trajectory descriptor encodes the shape of the trajectory represented by the normalized relative coor- × dinates of the successive points forming the trajectory. It directly depends on the dense flow used for tracking points. 2. HOF is computed using the orientations and magnitudes of the flow field. 3. MBH is designed to capture the gradient of horizontal and vertical components of the flow. The motion boundaries encode the relative pixel motion and therefore suppress camera motion, but only to some extent. 4. HOG encodes the appearance by using the intensity gradient orientations and magnitudes. It is formally not a motion descriptor. Yet the position where the descriptor is computed depends on the trajectory shape. As in [29], volume around a feature point is divided into a 2 2 3 space-time grid. The orientations are quantized ian 2to × ×8 b2i ×ns 3fo srp HacOe-Gti amned g g9r ibdi.ns T fhoer o oHriOenFt (awtioitnhs one a qdudainttiiozneadl zero bin). The horizontal and vertical components of MBH are separately quantized into 8 bins each. 4.2. Impact of motion compensation The optical flow is simply referred to as flow in the following, while the compensated flow (see subsection 2. 1) is denoted by ω-flow. Both of them are considered in the tracking and descriptor computation stages. The trajectories obtained by tracking with the ω-flow are called ω-trajectories. Figure 2 comparatively illustrates the ωtrajectories and the trajectories obtained using the flow. The input video shows a man moving away from the car. In this video excerpt, the camera is following the man walking to the right, thus inducing a global motion to the left in the video. When using the flow, the computed trajectories reflect the combination of these two motion components (camera and scene motion) as depicted by Subfigure 2(b), which hampers the characterization of the current action. In contrast, the ω-trajectories plotted in Subfigure 2(c) are more active on the actor moving on the foreground, while those localized in the background are now parallel to the time axis enhancing static parts of the scene. The ω-trajectories are therefore more relevant for action recognition, since they are more regularly and more exclusively following the actor’s motion. Impact on Trajectory and HOG descriptors. Table 1reports the impact of ω-trajectories on Trajectory and HOG descriptors, which are both significantly improved by 3%4% of mAP on the two datasets. When improved by ωflow, these descriptors will be respectively referred to as ω-Trajdesc and ω-HOG in the rest of the paper. Although the better performance of ω-Trajdesc versus the original Trajectory descriptor was expected, the one 222555555866 2. Trajectories obtained from optical and compensated flows. The green tail is the trajectory the current frame. The trajectories are sub-sampled for the sake of clarity. The frames are extracted Figure over every 15 frames with red dot indicating 5 frames in this example. DescriptorHollywood2HMDB51 BaseTrliaωnje- Tc(rtoarejrdpyreos[c2d9u]ced)54 7 1. 7 4% %2382.–89% BaseliHnωOe- (GHreOp [2rG9od]uced)4 451 . 658%%%2296.– 13%% Table 1. ω-Trajdesc and ω-HOG: Impact of compensating flow on Trajectory descriptor and HOG descriptors. achieved by ω-HOG might be surprising. Our interpretation is that HOG captures more context with the modified trajectories. More precisely, the original HOG descriptor is computed from a 2D+t sub-volume aligned with the corresponding trajectory and hence represents the appearance along the trajectory shape. When using ω-flow, we do not align the video sequence. As a result, the ω-HOG descriptor is no more computed around the very same tracked physical point in the space-time volume but around points lying in a patch of the initial feature point, whose size depends on the affine flow magnitude. ω-HOG can be viewed as a “patchbased” computation capturing more information about the appearance of the background or of the moving foreground. As for ω-trajectories, they are closer to the real trajectories of the moving actors as they usually cancel the camera movement, and so, more easier to train and recognize. Impact on HOF. The ω-flow impacts computation used as an input to HOF computation itself. Therefore, HOF can both types of trajectories (ω-trajectories both the trajectory and the descriptor be computed along or those extracted MethodHollywood2HMDB51 Table(ω2rHf.-alocO IwomkF)inpHgacOtFobf[2u9ωhsb]i:f-nlo ωgwot-hwωHOflFown5H 02 34O. 58291F% %descripto3 r706s38.:–1076% m%APfor Hollywood2 and average accuracy for HMDB5 1. The ω-HOF is used in subsequent evaluations. from flow) and can encode both kinds of flows (ω-flow or flow). For the sake of completeness, we evaluate all the variants as well as the combination of both flows in the descriptor computation stage. The results are presented in Table 2 and demonstrate the significant improvement obtained by computing the HOF descriptor with the ω-flow instead of the optical flow. Note that the type of trajectories which is used, either “Tracking flow” or “Tracking ω-flow”, has a limited impact in this case. From now on, we only consider the “Tracking ω-flow” case where HOF is computed along ω-trajectories. Interestingly, combining the HOF computed from the flow and the ω-flow further improves the results. This suggests that the two flow fields are complementary and the affine flow that was subtracted from ω-flow brings in additional information. For the sake of brevity, the combination of the two kinds of HOF, i.e., computed from the flow and the ω-flow using ω-trajectories, is referred to as the ω-HOF 222555555977 MethodHollywood2HMDB51 Tab(lerT3a.cIkmM inpBgMacHgωtBf-loH w [u2)s9in]gω f-lo wo MBH5 d42 e.052s7c% riptos:m34A90P.–3769f% orHllywood2 and average accuracy for HMDB5 1. DTerHasMjcBrOeblitpHGeFor4.ySumTωraw- frcilykto hw ionfgtheduωpCs-fcaolrtωmeiwNp-df/tl+Aωoutrw-finlogwthdesωcr- isTpc-fHtrMloaiOjrBpdswtGeHFosrc descriptor in the rest of this paper. Compared to the HOF baseline, the ω-HOF descriptor achieves a gain of +3.1% of mAP on Hollywood 2 and of +7.8% on HMDB51. Impact on MBH. Since MBH is computed from gradient of flow and cancel the constant motion, there is practically no benefit in using the ω-flow to compute the MBH descriptors, as shown in Table 3. However, by tracking ω-flow, the performance improves by around 1.3% for HMDB5 1 dataset and drops by around 1.5% for Hollywood2. This relative performance depends on the encoding technique. We will come back on this descriptor when considering another encoding scheme for local descriptors in Section 6. 4.3. Summary of compensated descriptors Table 4 summarizes the refined versions of the descriptors obtained by exploiting the ω-flow, and both ω-flow and the optical flow in the case of HOF. The revisited descriptors considerably improve the results compared to the orig- inal ones, with the noticeable exception of ω-MBH which gives mixed performance with a bag-of-features encoding scheme. But we already mention as this point that this incongruous behavior of ω-MBH is stabilized with the VLAD encoding scheme considered in Section 6. Another advantage of tracking the compensated flow is that fewer trajectories are produced. For instance, the total number of trajectories decreases by about 9. 16% and 22.81% on the Hollywood2 and HMDB51 datasets, respectively. Note that exploiting both the flow and the ω-flow do not induce much computational overhead, as the latter is obtained from the flow and the affine flow which is computed in real-time and already used to get the ω-trajectories. The only additional computational cost that we introduce by using the descriptors summarized in Table 4 is the computation of a second HOF descriptor, but this stage is relatively efficient and not the bottleneck of the extraction procedure. 5. Divergence-Curl-Shear descriptor This section introduces a new descriptor encoding the kinematic properties of motion discussed in Section 2.2. It is denoted by DCS in the rest of this paper. Combining kinematic features. The spatial derivatives are computed for the horizontal and vertical components of the flow field, which are used in turn to compute the divergence, curl and shear scalar values, see Equation 3. We consider all possible pairs of kinematic features, namely (div, curl), (div, shear) and (curl, shear). At each × ×× pixel, we compute the orientation and magnitude of the 2-D vector corresponding to each of these pairs. The orientation is quantized into histograms and the magnitude is used for weighting, similar to SIFT. Our motivation for encoding pairs is that the joint distribution of kinematic features conveys more information than exploiting them independently. Implementation details. The descriptor computation and parameters are similar to HOG and other popular descriptors such as MBH, HOF. We obtain 8-bin histograms for each of the three feature pairs or components of DCS. The range of possible angles is 2π for the (div,curl) pair and π for the other pairs, because the shear is always positive. The DCS descriptor is computed for a space-time volume aligned with a trajectory, as done with the four descriptors mentioned in the previous section. In order to capture the spatio-temporal structure of kinematic features, the volume (32 32 pixels and L = 15 frames) is subdivided into a spatio-temporal grid nofd s Lize = nx 5× f ny m×e nt, sw situhb nx =de ny =to 2a and nt = 3. These parameters ×hnave× × bneen fixed for the sake of consistency with the other descriptors. For each pair of kinematic features, each cell in the grid is represented by a histogram. The resulting local descriptors have a dimensionality equal to 288 = nx ny nt 8 3. At the video level, these descriptors are nenc×od end i×nto 8 a single vector representation using either BOF or the VLAD encoding scheme introduced in the next section. 6. VLAD in actions VLAD [8] is a descriptor encoding technique that aggregates the descriptors based on a locality criterion in the feature space. To our knowledge, this technique has never been considered for action recognition. Below, we briefly introduce this approach and give the performance achieved for all the descriptors introduced along the previous sections. VLAD in brief. Similar to BOF, VLAD relies on a codebook C = {c1, c2 , ...ck} of k centroids learned by k-means. bTohoek representation is ob}t oaifn ked c by summing, efodr b yea kch-m mveiasunasl. word ci, the differences x − ci of the vectors x assigned to ci, thereby producing a sv exct −or c representation oflength d×k, 222555556088 DMeBscHriptorV5 LH.A1o%Dlywo5Bo4d.O2 %F4V3L.3HA%MD B35B91.O7%F Taωbl-eDHM5rOCBa.FSjGPdHe+rsωfco-mMHaBOnFeofV54L2936A.51D% with5431ω208-.5T96% rajde3s42c97158,.ω3% -HOG342,58019ω.6-% HOF descriptors and their combination. where d is the dimension ofthe local descriptors. We use the codebook size, k = 256. Despite this large dimensionality, VLAD is efficient because it is effectively compared with a linear kernel. VLAD is post-processed using a componentwise power normalization, which dramatically improves its performance [8]. While cross validating the parameter α involved in this power normalization, we consistently observe, for all the descriptors, a value between 0.15 and 0.3. Therefore, this parameter is set to α = 0.2 in all our experiments. For classification, we use a linear SVM and oneagainst-rest approach everywhere, unless stated otherwise. Impact on existing descriptors. We employ VLAD because it is less sensitive to quantization parameters and appears to provide better performance with descriptors having a large dimensionality. These properties are interesting in our case, because the quantization parameters involved in the DCS and MBH descriptors have been used unchanged in Section 4 for the sake of direct comparison. They might be suboptimal when using the ω-flow instead of the optical flow on which they have initially been optimized [29]. Results for MBH and ω-MBH in Table 5 supports this argument. When using VLAD instead of BOF, the scores are stable in both the cases and there is no mixed inference as that observed in Table 3. VLAD also has significant positive influence on accuracy of ω-DCS descriptor. We also observe that ω-DCS is complementary to ω-MBH and adds to the performance. Still DCS is probably not best utilized in the current setting of parameters. In case of ω-Trajdesc and ω-HOG, the scores are better with BOF on both the datasets. ω-HOF with VLAD improves on HMDB5 1, but remains equivalent for Hollywood2. Although BOF leads to better scores for the descriptors considered individually, their combination with VLAD outperforms the BOF. 7. Comparison with the state of the art This section reports our results with all descriptors combined and compares our method with the state of the art. TrajectorCy+omHbOiGna+tHioOnF+ MDBCHSHol5 l98y.w76%o%od2H4M489.D02%B%51 All ω-descriptors all five compensated descriptors using combined62.5%52.1% Table 6. Combination of VLAD representation. WU*JVliaOnughreM tHaeolth. [yo2w9d87o] 256 0985. 37% SKa*duOeJinhrau tnegdMteHatlMa.h [ol1Dd.0B[91]25 24 609.8172% Table 7. Comparison with the state of the art on Hollywood2 and HMDB5 1 datasets. *Vig et al. [28] gets 61.9% by using external eye movements data. *Jiang et al. [9] used one-vs-one multi class SVM while our and other methods use one-vs-rest SVMs. With one-against-one multi class SVM we obtain 45. 1% for HMDB51. Descriptor combination. Table 6 reports the results obtained when the descriptors are combined. Since we use VLAD, our baseline is updated that is combination of Trajectory, HOG, HOF and MBH with VLAD representation. When DCS is added to the baseline there is an improvement of 0.9% and 1.2%. With combination of all five compensated descriptors we obtain 62.5% and 52.1% on the two datasets. This is a large improvement even over the updated baseline, which shows that the proposed motion compensation and the way we exploit it are significantly important for action recognition. The comparison with the state of the art is shown in Table 7. Our method outperforms all the previously reported results in the literature. In particular, on the HMDB51 dataset, the improvement over the best reported results to date is more than 11% in average accuracy. Jiang el al. [9] used a one-against-one multi-class SVM, which might have resulted in inferior scores. With a similar multi-class SVM approach, our method obtains 45. 1%, which remains significantly better than their result. All others results were reported with one-against-rest approach. On Olympic Sports dataset we obtain mAP of 83.2% with ‘All ω-descriptors combined’ and the improvement is mostly because of VLAD and ω-flow. The best reported mAPs on this dataset are Liu et al. [14] (74.4%) and Jiang et al. [9] (80.6%), which we exceed convincingly. Gaidon et al. [6] reports the best average accuracy of 82.7%. 8. Conclusions This paper first demonstrates the interest of canceling the dominant motion (predominantly camera motion) to make the visual motion truly related to actions, for both the trajectory extraction and descriptor computation stages. It pro222555556199 duces significantly better versions (called compensated descriptors) of several state-of-the-art local descriptors for action recognition. The simplicity, efficiency and effectiveness of this motion compensation approach make it applicable to any action recognition framework based on motion descriptors and trajectories. The second contribution is the new DCS descriptor derived from the first-order scalar motion quantities specifying the local motion patterns. It captures additional information which is proved complementary to the other descriptors. Finally, we show that VLAD encoding technique instead of bag-of-words boosts several action descriptors, and overall exhibits a significantly better performance when combining different types of descriptors. Our contributions are all complementary and significantly outperform the state of the art when combined, as demon- strated by our extensive experiments on the Hollywood 2, HMDB51 and Olympic Sports datasets. Acknowledgments This work was supported by the Quaero project, funded by Oseo, French agency for innovation. We acknowledge Heng Wang’s help for reproducing some of their results. References [1] S. Ali and M. Shah. Human action recognition in videos using kinematic features and multiple instance learning. IEEE T-PAMI, 32(2):288–303, Feb. 2010. [2] T. Brox and J. Malik. Object segmentation by long term analysis of point trajectories. In ECCV, Sep. 2010. [3] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In CVPR, Jun. 2005. [4] N. Dalal, B. Triggs, and C. Schmid. Human detection using oriented histograms of flow and appearance. In ECCV, May 2006. [5] P. Dollar, V. Rabaud, G. Cottrell, and S. Belongie. Behavior recognition via sparse spatio-temporal features. In VS-PETS, Oct. 2005. [6] A. Gaidon, Z. Harchaoui, and C. Schmid. Recognizing activities with cluster-trees of tracklets. In BMVC, Sep. 2012. [7] A. Hervieu, P. Bouthemy, and J.-P. Le Cadre. A statistical video content recognition method using invariant features on object trajectories. IEEE T-CSVT, 18(1 1): 1533–1543, 2008. [8] H. J ´egou, F. Perronnin, M. Douze, J. S ´anchez, P. P ´erez, and C. Schmid. Aggregating local descriptors into compact [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] codes. IEEE T-PAMI, 34(9):1704–1716, 2012. Y.-G. Jiang, Q. Dai, X. Xue, W. Liu, and C.-W. Ngo. Trajectory-based modeling of human actions with motion reference points. In ECCV, Oct. 2012. O. Kliper-Gross, Y. Gurovich, T. Hassner, and L. Wolf. Motion interchange patterns for action recognition in unconstrained videos. In ECCV, Oct. 2012. H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. Hmdb: A large video database for human motion recognition. In ICCV, Nov. 2011. I. Laptev and T. Lindeberg. Space-time interest points. In ICCV, Oct. 2003. I. Laptev, M. Marzalek, C. Schmid, and B. Rozenfeld. Learning realistic human actions from movies. In CVPR, Jun. 2008. J. Liu, B. Kuipers, and S. Savarese. Recognizing human actions by attributes. In CVPR, Jun. 2011. D. Lowe. Distinctive image features from scale-invariant keypoints. IJCV, 60(2):91–1 10, Nov. 2004. M. Marzalek, I. Laptev, and C. Schmid. Actions in context. In CVPR, Jun. 2009. P. Matikainen, M. Hebert, and R. Sukthankar. Trajectons: Action recognition through the motion analysis of tracked features. In Workshop on Video-Oriented Object and Event Classification, ICCV, Sep. 2009. R. Messing, C. J. Pal, and H. A. Kautz. Activity recognition using the velocity histories of tracked keypoints. In ICCV, Sep. 2009. J. C. Niebles, C.-W. Chen, and F.-F. Li. Modeling temporal structure of decomposable motion segments for activity classification. In ECCV, Sep. 2010. [20] J.-M. Odobez and P. Bouthemy. Robust multiresolution estimation of parametric motion models. Jal of Vis. Comm. and Image Representation, 6(4):348–365, Dec. 1995. [21] G. Piriou, P. Bouthemy, and J.-F. Yao. Recognition of dynamic video contents with global probabilistic models of visual motion. IEEE T-IP, 15(1 1):3417–3430, 2006. [22] S. Sadanand and J. J. Corso. Action bank: A high-level representation of activity in video. In CVPR, Jun. 2012. [23] C. Schmid and R. Mohr. Local grayvalue invariants for image retrieval. IEEE T-PAMI, 19(5):530–534, May 1997. [24] J. Sivic and A. Zisserman. Video Google: A text retrieval approach to object matching in videos. In ICCV, pages 1470– 1477, Oct. 2003. [25] J. Sun, X. Wu, S. Yan, L. F. Cheong, T.-S. Chua, and J. Li. Hierarchical spatio-temporal context modeling for action recognition. In CVPR, Jun. 2009. [26] H. Uemura, S. Ishikawa, and K. Mikolajczyk. Feature tracking and motion compensation for action recognition. In BMVC, Sep. 2008. [27] M. M. Ullah, S. N. Parizi, and I. Laptev. Improving bag-offeatures action recognition with non-local cues. In BMVC, Sep. 2010. [28] E. Vig, M. Dorr, and D. Cox. Saliency-based space-variant descriptor sampling for action recognition. In ECCV, Oct. 2012. [29] H. Wang, A. Kl¨ aser, C. Schmid, and C.-L. Liu. Action recognition by dense trajectories. In CVPR, Jun. 2011. [30] H. Wang, M. M. Ullah, A. Kl¨ aser, I. Laptev, and C. Schmid. Evaluation of local spatio-temporal features for action recognition. In BMVC, Sep. 2009. [3 1] G. Willems, T. Tuytelaars, and L. J. V. Gool. An efficient dense and scale-invariant spatio-temporal interest point detector. In ECCV, Oct. 2008. [32] S. Wu, O. Oreifej, and M. Shah. Action recognition in videos acquired by a moving camera using motion decomposition of lagrangian particle trajectories. In ICCV, Nov. 2011. 222555666200
2 0.99932605 137 cvpr-2013-Dynamic Scene Classification: Learning Motion Descriptors with Slow Features Analysis
Author: Christian Thériault, Nicolas Thome, Matthieu Cord
Abstract: In this paper, we address the challenging problem of categorizing video sequences composed of dynamic natural scenes. Contrarily to previous methods that rely on handcrafted descriptors, we propose here to represent videos using unsupervised learning of motion features. Our method encompasses three main contributions: 1) Based on the Slow Feature Analysis principle, we introduce a learned local motion descriptor which represents the principal and more stable motion components of training videos. 2) We integrate our local motion feature into a global coding/pooling architecture in order to provide an effective signature for each video sequence. 3) We report state of the art classification performances on two challenging natural scenes data sets. In particular, an outstanding improvement of 11 % in classification score is reached on a data set introduced in 2012.
3 0.99922204 260 cvpr-2013-Learning and Calibrating Per-Location Classifiers for Visual Place Recognition
Author: Petr Gronát, Guillaume Obozinski, Josef Sivic, Tomáš Pajdla
Abstract: The aim of this work is to localize a query photograph by finding other images depicting the same place in a large geotagged image database. This is a challenging task due to changes in viewpoint, imaging conditions and the large size of the image database. The contribution of this work is two-fold. First, we cast the place recognition problem as a classification task and use the available geotags to train a classifier for each location in the database in a similar manner to per-exemplar SVMs in object recognition. Second, as onlyfewpositive training examples are availablefor each location, we propose a new approach to calibrate all the per-location SVM classifiers using only the negative examples. The calibration we propose relies on a significance measure essentially equivalent to the p-values classically used in statistical hypothesis testing. Experiments are performed on a database of 25,000 geotagged street view images of Pittsburgh and demonstrate improved place recognition accuracy of the proposed approach over the previous work. 2Center for Machine Perception, Faculty of Electrical Engineering 3WILLOW project, Laboratoire d’Informatique de l’E´cole Normale Sup e´rieure, ENS/INRIA/CNRS UMR 8548. 4Universit Paris-Est, LIGM (UMR CNRS 8049), Center for Visual Computing, Ecole des Ponts - ParisTech, 77455 Marne-la-Valle, France
4 0.99917489 346 cvpr-2013-Real-Time No-Reference Image Quality Assessment Based on Filter Learning
Author: Peng Ye, Jayant Kumar, Le Kang, David Doermann
Abstract: This paper addresses the problem of general-purpose No-Reference Image Quality Assessment (NR-IQA) with the goal ofdeveloping a real-time, cross-domain model that can predict the quality of distorted images without prior knowledge of non-distorted reference images and types of distortions present in these images. The contributions of our work are two-fold: first, the proposed method is highly efficient. NR-IQA measures are often used in real-time imaging or communication systems, therefore it is important to have a fast NR-IQA algorithm that can be used in these real-time applications. Second, the proposed method has the potential to be used in multiple image domains. Previous work on NR-IQA focus primarily on predicting quality of natural scene image with respect to human perception, yet, in other image domains, the final receiver of a digital image may not be a human. The proposed method consists of the following components: (1) a local feature extractor; (2) a global feature extractor and (3) a regression model. While previous approaches usually treat local feature extraction and regres- sion model training independently, we propose a supervised method based on back-projection, which links the two steps by learning a compact set of filters which can be applied to local image patches to obtain discriminative local features. Using a small set of filters, the proposed method is extremely fast. We have tested this method on various natural scene and document image datasets and obtained stateof-the-art results.
5 0.99904537 165 cvpr-2013-Fast Energy Minimization Using Learned State Filters
Author: Matthieu Guillaumin, Luc Van_Gool, Vittorio Ferrari
Abstract: Pairwise discrete energies defined over graphs are ubiquitous in computer vision. Many algorithms have been proposed to minimize such energies, often concentrating on sparse graph topologies or specialized classes of pairwise potentials. However, when the graph is fully connected and the pairwise potentials are arbitrary, the complexity of even approximate minimization algorithms such as TRW-S grows quadratically both in the number of nodes and in the number of states a node can take. Moreover, recent applications are using more and more computationally expensive pairwise potentials. These factors make it very hard to employ fully connected models. In this paper we propose a novel, generic algorithm to approximately minimize any discrete pairwise energy function. Our method exploits tractable sub-energies to filter the domain of the function. The parameters of the filter are learnt from instances of the same class of energies with good candidate solutions. Compared to existing methods, it efficiently handles fully connected graphs, with many states per node, and arbitrary pairwise potentials, which might be expensive to compute. We demonstrate experimentally on two applications that our algorithm is much more efficient than other generic minimization algorithms such as TRW-S, while returning essentially identical solutions.
6 0.99880189 180 cvpr-2013-Fully-Connected CRFs with Non-Parametric Pairwise Potential
7 0.99880177 55 cvpr-2013-Background Modeling Based on Bidirectional Analysis
8 0.99836129 93 cvpr-2013-Constraints as Features
9 0.99826252 113 cvpr-2013-Dense Variational Reconstruction of Non-rigid Surfaces from Monocular Video
10 0.99820435 178 cvpr-2013-From Local Similarity to Global Coding: An Application to Image Classification
11 0.9981941 48 cvpr-2013-Attribute-Based Detection of Unfamiliar Classes with Humans in the Loop
12 0.99810183 357 cvpr-2013-Revisiting Depth Layers from Occlusions
13 0.99755234 301 cvpr-2013-Multi-target Tracking by Rank-1 Tensor Approximation
14 0.99690282 252 cvpr-2013-Learning Locally-Adaptive Decision Functions for Person Verification
15 0.99391073 189 cvpr-2013-Graph-Based Discriminative Learning for Location Recognition
16 0.99305809 378 cvpr-2013-Sampling Strategies for Real-Time Action Recognition
17 0.99287128 379 cvpr-2013-Scalable Sparse Subspace Clustering
18 0.99245805 343 cvpr-2013-Query Adaptive Similarity for Large Scale Object Retrieval
19 0.99200439 266 cvpr-2013-Learning without Human Scores for Blind Image Quality Assessment
20 0.99170727 148 cvpr-2013-Ensemble Video Object Cut in Highly Dynamic Scenes