cvpr cvpr2013 cvpr2013-315 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Cewu Lu, Jiaping Shi, Jiaya Jia
Abstract: Online dictionary learning is particularly useful for processing large-scale and dynamic data in computer vision. It, however, faces the major difficulty to incorporate robust functions, rather than the square data fitting term, to handle outliers in training data. In thispaper, wepropose a new online framework enabling the use of ?1 sparse data fitting term in robust dictionary learning, notably enhancing the usability and practicality of this important technique. Extensive experiments have been carried out to validate our new framework.
Reference: text
sentIndex sentText sentNum sentScore
1 hk Abstract Online dictionary learning is particularly useful for processing large-scale and dynamic data in computer vision. [sent-4, score-0.652]
2 It, however, faces the major difficulty to incorporate robust functions, rather than the square data fitting term, to handle outliers in training data. [sent-5, score-0.254]
3 In thispaper, wepropose a new online framework enabling the use of ? [sent-6, score-0.353]
4 1 sparse data fitting term in robust dictionary learning, notably enhancing the usability and practicality of this important technique. [sent-7, score-0.859]
5 Sparse coding and dictionary learning have been widely employed in many computer vision tasks, including image denoising [2], texture synthesis [10], and image inpainting [8], as well as higher-level tasks such as image classification [17] and unusual event detection [18]. [sent-11, score-0.732]
6 A general framework for sparse dictionary learning is to minimize the objective function Q(Bn,D) =n1i? [sent-12, score-0.68]
7 In sparse representation, each xi is a sparse linear combination over a set of basis vectors from an over-complete dictionary D in Rp×q with its corresponding sparse coefficient βi. [sent-20, score-0.79]
8 Recent online strategies [8, 14] only update one βi for xi each time in Bn. [sent-24, score-0.512]
9 It is shown in [8] that this type of online solvers can achieve comparable performance as the batch processing scheme. [sent-25, score-0.69]
10 We denote Θb as the batch dictionary estimator and as the online dictionary estimator. [sent-27, score-1.868]
11 The two types of dictionary learning are respectively expressed as Θo Dbnatch = Θb(xn, Xn−1), (2) Donnline = Θo(xn, Dn−1). [sent-28, score-0.66]
12 (3) In the online process, there is no need to record and process Xn in each round. [sent-29, score-0.353]
13 Thus, linearly increased computation and constant memory consumption are resulted in as data size grows. [sent-31, score-0.149]
14 Online dictionary learning [8, 14, 13] is vastly useful for applications that involve large-scale and dynamic data, such as video sequences [11], for which traditional batch process is not suitable. [sent-32, score-1.022]
15 All existing online dictionary learning methods do not use the robust function in the data fitting term and might be vulnerable to large outliers. [sent-33, score-1.194]
16 This apparently simple form, unfortunately, causes tremen- dous trouble for online processing because μn cannot be quickly estimated given only the history result μn−1 = median(x1 , . [sent-51, score-0.419]
17 So even with the 444111335 computed μn−1, finding μn may still require long computation time, losing all benefits brought by the online scheme. [sent-56, score-0.353]
18 =1(xi− μ)2, (5) online parameter estimator can be easily written as μn = nn−1μn−1 + n1xn. [sent-62, score-0.353]
19 Our Contribution We propose an online robust dictionary learning (ORDL) algorithm. [sent-69, score-1.053]
20 Our ORDL approach takes the full advantage of an online scheme that is, it consumes constant memory as data grow and takes time linear to the size of data. [sent-71, score-0.469]
21 In the meantime, it is very robust against outliers thanks to the sparsity nature. [sent-72, score-0.157]
22 Our method makes dictionary learning more practical and usable in many computer vision applications, exemplified in this paper. [sent-73, score-0.631]
23 , xn}, where each xi is normalized as zero mean and unit variance, we assume that dictionary D belongs to a closed, convex, and bounded set D as D ? [sent-78, score-0.643]
24 Previous online dictionary learning [8, 14] is based on the function ∈ Rp×q s. [sent-80, score-0.984]
25 Different from it, our robust dictionary learning is defined as Dm∈Din,B1ni? [sent-96, score-0.7]
26 ,xn}, regularization parameter λ, initial dictionary D0, mini-batch data size h 2: for each {xt−h+1 , . [sent-108, score-0.589]
27 By tradition, online dictionary learning updates D and robust sparse coding estimates B, which is achieved by iterative reweighed least squares discussed in [19]. [sent-121, score-1.151]
28 Batch Dictionary Update We begin with the batch dictionary update procedure and then extend it to the online version. [sent-128, score-1.384]
29 Theoretical robustness analysis of our online dictionary update is provided. [sent-129, score-1.073]
30 In batch robust dictionary learning [19], updating of the dictionary assumes that all elements in B are provided. [sent-130, score-1.674]
31 , βn] ∈ Rq×n in order to distinguish them from those in the online procedure. [sent-137, score-0.353]
32 Batch dictionary updating is accordingly expressed as mDinn1? [sent-138, score-0.666]
33 =n1|xij− dβi|, where d ∈ R1 ×q and xij (10) is the jth element of xi. [sent-149, score-0.177]
34 The IRLS scheme solves the following two problem in each iteration 444111446 Algorithm 2 Batch Dictionary Update 1:input: x1,··· ,xn; β1,··· ,βn; random matrix D; wij = 1for all i= 1, · · · , n, j = 1, · · · ,p 2: repeat 3: for j = 1 : p do 4: Mj = ? [sent-150, score-0.175]
35 in=1 wijxijβiT 5: 6: solve linear system Cj = D(j, :)Mj 7: calculate wij (i = 1, 2, . [sent-152, score-0.173]
36 Online Dictionary Update Extending the above process to online is not straightforward because we only see new mini-batch data xt−h+1 , . [sent-173, score-0.353]
37 where = is With this configuration, we update Dt using Algorithm 3 in an online style. [sent-192, score-0.458]
38 ,p 2: repeat 3: for j = 1 : p do wij 4: Mtj ← Mtj−h +? [sent-205, score-0.15]
39 tk=t−h+1wijxijβiT solve linear system Ctj = D(j, :)Mtj 5: 6: 7: calculate wij (i = t − h + 1, . [sent-207, score-0.173]
40 , t − h), all wij = 1 2: repeat 3: calculate d by Eq. [sent-221, score-0.173]
41 (12) 5: until converge 6: output: Dt = D, wij Mtj, Ctj previous Dt−h as initialization in the current round. [sent-226, score-0.173]
42 In practice, thanks to the warm start of Dt−h, dictionary update in Algorithm 3 converges in 5 ∼ 7 iterations. [sent-231, score-0.716]
43 Analysis We analyze how robust the online algorithm is. [sent-234, score-0.422]
44 We demonstrate that updating the jth row of dictionary in Algorithm 3 is equivalent to solving t− h t mdin? [sent-235, score-0.683]
45 Therefore, estimation of each row of the dictionary in Algorithm 3 is achieved by solving Eq. [sent-245, score-0.589]
46 According to robust statistics [12], high quality results can be obtained if the number of outliers in the coming data is smaller than h/2. [sent-254, score-0.157]
47 An outlier sample xij will find a small corresponding weight according to Eq. [sent-255, score-0.203]
48 As training data grow, the history term carries inlier information, in turn enhancing outlier rejection by Eq. [sent-258, score-0.219]
49 Connection between Online and Batch Schemes The online method seemingly differs from the batch version. [sent-264, score-0.69]
50 To understand it, we rewrite the batch update as t− h t mdin? [sent-266, score-0.442]
51 Online robust dictionary learning (ORDL) has a weighted square, i. [sent-271, score-0.7]
52 In comparison, batch robust dictionary learning (BRDL) contains the ? [sent-275, score-1.037]
53 Experiments We qualitatively and quantitatively evaluate our online framework on different sets of data and applications. [sent-282, score-0.353]
54 There has been research [19] showing that robust dictionary learning incorporating the ? [sent-283, score-0.7]
55 We show more evidences, especially using the online scheme. [sent-286, score-0.353]
56 Evaluation on Synthetic Data The performance of online robust dictionary update is compared with that of the batch process. [sent-292, score-1.453]
57 Given a dictionary D ∈ R32× 100, we generate training data xi = Dβi ξi (1 ≤ i ≤ n), where ξi is additive Laplace noise to simulate outliers. [sent-319, score-0.666]
58 We measure the difference between the batch dictionary result Db and the online one Do via correlation Corr(Db , Do). [sent-322, score-1.279]
59 We also evaluate how our online scheme updates the resulting dictionary by measuring the dictionary difference ? [sent-327, score-1.556]
60 We also compare running time and memory consumption and plot them in Fig. [sent-333, score-0.182]
61 We conclude that our online dictionary update uses linear time and nearly constant memory. [sent-335, score-1.047]
62 The batch process, on the contrary, takes polynomial + λ 444111668 Figure 3. [sent-336, score-0.337]
63 Image inpainting results by online and batch dictionary learning. [sent-337, score-1.328]
64 We carry out another synthetic experiment to evaluate the robustness of our online method. [sent-343, score-0.379]
65 If a learning method can reject outlier in the training data, the total reconstruction error should be small. [sent-354, score-0.158]
66 It shows that the performance of our online robust dictionary learning (ORDL) is comparable to batch robust dictionary learning (BRDL) and outperforms those using the ? [sent-356, score-2.09]
67 Evaluation on Natural Image Inpainting We evaluate our online robust dictionary learning using color image inpainting, which aims to infer missing pixels due to superimposed text, subtitles, or watermark. [sent-360, score-1.053]
68 In this application, the dictionary is trained on natural image datasets [9] with 15% random outliers. [sent-361, score-0.589]
69 Basically, the batch and online processes produce very similar results; but the online )Ted(cionsme2130 0 BOR2SDLignal4Num6berx8104emtyBroeM)(109871. [sent-373, score-1.043]
70 For comparison, considering the dictionary update step only, the online procedure takes 0. [sent-381, score-1.047]
71 In the batch mode, it takes about 3000 seconds in total. [sent-383, score-0.337]
72 We report the time and memory consumption in Fig. [sent-384, score-0.149]
73 A few learned dictionary atoms using our Online Robust Dictionary Learning (ORDL). [sent-391, score-0.624]
74 Our experiments are carried out on handwritten digit datasets MNIST [6] and USPS [1]. [sent-398, score-0.171]
75 To verify the robustness of our approach, we add random outliers to the images (both training and testing data) and colorize them to greatly increase the versatility and complexity given the digits all in grayscale originally. [sent-401, score-0.177]
76 We learn separate dictionaries for each digit via outlier contaminated training data. [sent-408, score-0.288]
77 In the testing phase, given a digit image, we reconstruct it over the ten learned dictionaries and choose the one with smallest reconstruction error. [sent-409, score-0.154]
78 We compare our online robust learning (ORDL) with batch robust learning (BRDL) and other state-of-theart ? [sent-410, score-0.912]
79 2-norm data term dictionary learning methods, including KSVD [2] and Mairal et al. [sent-411, score-0.664]
80 The learned dictionary by our method is shown in Fig. [sent-413, score-0.589]
81 Background subtraction by (a) K-SVD [2] (b) batch robust dictionary update, and (c) our method with online dictionary update. [sent-433, score-1.977]
82 Our results do not contain the “ghost” artifacts, similar to those produced by batch dictionary learning. [sent-434, score-0.926]
83 That is, given the ith frame xi, its background is assumed to be sparsely represented as a linear combination of a few atoms in the learned dictionary D. [sent-442, score-0.701]
84 Foreground is represented as xi Dβi, which is assumed to be sparse after background subtraction. [sent-444, score-0.148]
85 In our model, background subtraction can be achieved by solving our robust dictionary learning model − mβini? [sent-445, score-0.785]
86 However, it works in a batch mode, difficult for long-video-sequence processing. [sent-465, score-0.337]
87 (a) input frames; (b) subtracted background with batch dictionary update; (c) foreground in batch mode; (d) background with online dictionary update; (e) foreground by our online algorithm. [sent-471, score-2.722]
88 In our online dictionary update, new data (or min-bacth) are with size h = 20. [sent-472, score-0.942]
89 We compare batch and online results with those of K-SVD [2] in Figs. [sent-473, score-0.69]
90 Our online dictionary learning takes much less time than the batch scheme, as compared in Fig. [sent-476, score-1.321]
91 11, we compare the extracted foreground segments by our online dictionary learning method, the batch one [19], and by the method using the ? [sent-479, score-1.358]
92 it shows that our online robust dictionary learning method can achieve comparable accu- orTu)(smHei30142OBS1R0izD eLofT20rain30gDa4t05 )GyromeMB(210. [sent-484, score-1.053]
93 BRDL: batch robust dictionary learning; ORDL: online robust dictionary learning. [sent-489, score-2.006]
94 racy with batch robust dictionary learning and outperforms those ? [sent-490, score-1.037]
95 Conclusion We have proposed an online robust dictionary learning framework. [sent-495, score-1.053]
96 The online dictionary update scheme saves time and memory while not sacrificing much accuracy. [sent-496, score-1.172]
97 We have developed an algorithm with full details to accomplish it and have shown that this algorithm can produce similar-quality results as the batch robust one. [sent-497, score-0.429]
98 (a) inputs; (b) our results using online dictionary update; (c) results of [19] using an ? [sent-513, score-0.942]
99 1-norm data fitting term; (d) sparse coding and dictionary update with the ? [sent-514, score-0.846]
100 2-norm data fitting term; (e) ground truth foreground masks; (f)-(h) foreground masks corresponding to (b)-(d). [sent-515, score-0.148]
wordName wordTfidf (topN-words)
[('dictionary', 0.589), ('online', 0.353), ('batch', 0.337), ('mtj', 0.243), ('ordl', 0.199), ('ctj', 0.177), ('wij', 0.15), ('brdl', 0.133), ('xij', 0.131), ('digit', 0.124), ('wijxij', 0.111), ('update', 0.105), ('xn', 0.092), ('outliers', 0.088), ('db', 0.083), ('consumption', 0.078), ('fitting', 0.074), ('outlier', 0.072), ('memory', 0.071), ('robust', 0.069), ('irls', 0.069), ('history', 0.066), ('usps', 0.063), ('xt', 0.058), ('mnist', 0.055), ('xi', 0.054), ('dt', 0.051), ('inpainting', 0.049), ('sparse', 0.049), ('updating', 0.048), ('jth', 0.046), ('wji', 0.045), ('background', 0.045), ('mdin', 0.044), ('learning', 0.042), ('mairal', 0.042), ('ini', 0.04), ('subtraction', 0.04), ('digits', 0.04), ('contaminated', 0.039), ('foreground', 0.037), ('mj', 0.036), ('atoms', 0.035), ('ksvd', 0.034), ('vulnerable', 0.034), ('ghost', 0.034), ('rp', 0.033), ('term', 0.033), ('vastly', 0.033), ('running', 0.033), ('ith', 0.032), ('dictionaries', 0.03), ('saves', 0.029), ('reweighted', 0.029), ('expressed', 0.029), ('coding', 0.029), ('dn', 0.029), ('restored', 0.028), ('mode', 0.028), ('cj', 0.027), ('dot', 0.027), ('robustness', 0.026), ('handwritten', 0.025), ('enhancing', 0.025), ('scheme', 0.025), ('incoming', 0.024), ('records', 0.024), ('zhao', 0.024), ('unusual', 0.023), ('calculate', 0.023), ('accomplish', 0.023), ('vehicles', 0.023), ('traffic', 0.023), ('training', 0.023), ('converge', 0.023), ('coefficients', 0.022), ('converges', 0.022), ('carried', 0.022), ('bn', 0.022), ('reject', 0.021), ('overcomplete', 0.021), ('practically', 0.021), ('dynamic', 0.021), ('hong', 0.021), ('kong', 0.021), ('grow', 0.02), ('squares', 0.02), ('ptd', 0.02), ('poczos', 0.02), ('szabo', 0.02), ('correspondingto', 0.02), ('remedies', 0.02), ('substraction', 0.02), ('tradition', 0.02), ('jianping', 0.02), ('duarte', 0.02), ('usability', 0.02), ('baraniuk', 0.02), ('cevher', 0.02), ('posters', 0.02)]
simIndex simValue paperId paperTitle
same-paper 1 1.0000006 315 cvpr-2013-Online Robust Dictionary Learning
Author: Cewu Lu, Jiaping Shi, Jiaya Jia
Abstract: Online dictionary learning is particularly useful for processing large-scale and dynamic data in computer vision. It, however, faces the major difficulty to incorporate robust functions, rather than the square data fitting term, to handle outliers in training data. In thispaper, wepropose a new online framework enabling the use of ?1 sparse data fitting term in robust dictionary learning, notably enhancing the usability and practicality of this important technique. Extensive experiments have been carried out to validate our new framework.
2 0.43130642 296 cvpr-2013-Multi-level Discriminative Dictionary Learning towards Hierarchical Visual Categorization
Author: Li Shen, Shuhui Wang, Gang Sun, Shuqiang Jiang, Qingming Huang
Abstract: For the task of visual categorization, the learning model is expected to be endowed with discriminative visual feature representation and flexibilities in processing many categories. Many existing approaches are designed based on a flat category structure, or rely on a set of pre-computed visual features, hence may not be appreciated for dealing with large numbers of categories. In this paper, we propose a novel dictionary learning method by taking advantage of hierarchical category correlation. For each internode of the hierarchical category structure, a discriminative dictionary and a set of classification models are learnt for visual categorization, and the dictionaries in different layers are learnt to exploit the discriminative visual properties of different granularity. Moreover, the dictionaries in lower levels also inherit the dictionary of ancestor nodes, so that categories in lower levels are described with multi-scale visual information using our dictionary learning approach. Experiments on ImageNet object data subset and SUN397 scene dataset demonstrate that our approach achieves promising performance on data with large numbers of classes compared with some state-of-the-art methods, and is more efficient in processing large numbers of categories.
3 0.41087735 392 cvpr-2013-Separable Dictionary Learning
Author: Simon Hawe, Matthias Seibert, Martin Kleinsteuber
Abstract: Many techniques in computer vision, machine learning, and statistics rely on the fact that a signal of interest admits a sparse representation over some dictionary. Dictionaries are either available analytically, or can be learned from a suitable training set. While analytic dictionaries permit to capture the global structure of a signal and allow a fast implementation, learned dictionaries often perform better in applications as they are more adapted to the considered class of signals. In imagery, unfortunately, the numerical burden for (i) learning a dictionary and for (ii) employing the dictionary for reconstruction tasks only allows to deal with relatively small image patches that only capture local image information. The approach presented in this paper aims at overcoming these drawbacks by allowing a separable structure on the dictionary throughout the learning process. On the one hand, this permits larger patch-sizes for the learning phase, on the other hand, the dictionary is applied efficiently in reconstruction tasks. The learning procedure is based on optimizing over a product of spheres which updates the dictionary as a whole, thus enforces basic dictionary proper- , ties such as mutual coherence explicitly during the learning procedure. In the special case where no separable structure is enforced, our method competes with state-of-the-art dictionary learning methods like K-SVD.
4 0.39832661 257 cvpr-2013-Learning Structured Low-Rank Representations for Image Classification
Author: Yangmuzi Zhang, Zhuolin Jiang, Larry S. Davis
Abstract: An approach to learn a structured low-rank representation for image classification is presented. We use a supervised learning method to construct a discriminative and reconstructive dictionary. By introducing an ideal regularization term, we perform low-rank matrix recovery for contaminated training data from all categories simultaneously without losing structural information. A discriminative low-rank representation for images with respect to the constructed dictionary is obtained. With semantic structure information and strong identification capability, this representation is good for classification tasks even using a simple linear multi-classifier. Experimental results demonstrate the effectiveness of our approach.
5 0.38401625 66 cvpr-2013-Block and Group Regularized Sparse Modeling for Dictionary Learning
Author: Yu-Tseh Chi, Mohsen Ali, Ajit Rajwade, Jeffrey Ho
Abstract: This paper proposes a dictionary learning framework that combines the proposed block/group (BGSC) or reconstructed block/group (R-BGSC) sparse coding schemes with the novel Intra-block Coherence Suppression Dictionary Learning (ICS-DL) algorithm. An important and distinguishing feature of the proposed framework is that all dictionary blocks are trained simultaneously with respect to each data group while the intra-block coherence being explicitly minimized as an important objective. We provide both empirical evidence and heuristic support for this feature that can be considered as a direct consequence of incorporating both the group structure for the input data and the block structure for the dictionary in the learning process. The optimization problems for both the dictionary learning and sparse coding can be solved efficiently using block-gradient descent, and the details of the optimization algorithms are presented. We evaluate the proposed methods using well-known datasets, and favorable comparisons with state-of-the-art dictionary learning methods demonstrate the viability and validity of the proposed framework.
7 0.37305188 185 cvpr-2013-Generalized Domain-Adaptive Dictionaries
8 0.24839342 125 cvpr-2013-Dictionary Learning from Ambiguously Labeled Data
9 0.23676474 422 cvpr-2013-Tag Taxonomy Aware Dictionary Learning for Region Tagging
10 0.23378363 5 cvpr-2013-A Bayesian Approach to Multimodal Visual Dictionary Learning
11 0.20487985 302 cvpr-2013-Multi-task Sparse Learning with Beta Process Prior for Action Recognition
12 0.16291051 419 cvpr-2013-Subspace Interpolation via Dictionary Learning for Unsupervised Domain Adaptation
13 0.16246039 204 cvpr-2013-Histograms of Sparse Codes for Object Detection
14 0.14858888 142 cvpr-2013-Efficient Detector Adaptation for Object Detection in a Video
15 0.14182581 220 cvpr-2013-In Defense of Sparsity Based Face Recognition
16 0.1364347 399 cvpr-2013-Single-Sample Face Recognition with Image Corruption and Misalignment via Sparse Illumination Transfer
17 0.10804506 386 cvpr-2013-Self-Paced Learning for Long-Term Tracking
18 0.1060018 178 cvpr-2013-From Local Similarity to Global Coding: An Application to Image Classification
19 0.09608674 267 cvpr-2013-Least Soft-Threshold Squares Tracking
20 0.094901569 442 cvpr-2013-Transfer Sparse Coding for Robust Image Representation
topicId topicWeight
[(0, 0.175), (1, -0.175), (2, -0.307), (3, 0.351), (4, -0.128), (5, -0.148), (6, 0.172), (7, 0.115), (8, -0.014), (9, 0.152), (10, 0.002), (11, 0.053), (12, 0.005), (13, 0.053), (14, 0.019), (15, -0.013), (16, 0.024), (17, 0.038), (18, -0.002), (19, 0.008), (20, -0.01), (21, 0.042), (22, -0.012), (23, 0.002), (24, 0.021), (25, -0.049), (26, -0.034), (27, 0.034), (28, -0.034), (29, 0.057), (30, 0.073), (31, 0.012), (32, -0.041), (33, -0.009), (34, 0.015), (35, -0.021), (36, 0.013), (37, -0.032), (38, -0.038), (39, -0.011), (40, 0.042), (41, 0.011), (42, -0.022), (43, 0.054), (44, 0.007), (45, 0.005), (46, -0.036), (47, 0.013), (48, 0.066), (49, 0.029)]
simIndex simValue paperId paperTitle
same-paper 1 0.9768008 315 cvpr-2013-Online Robust Dictionary Learning
Author: Cewu Lu, Jiaping Shi, Jiaya Jia
Abstract: Online dictionary learning is particularly useful for processing large-scale and dynamic data in computer vision. It, however, faces the major difficulty to incorporate robust functions, rather than the square data fitting term, to handle outliers in training data. In thispaper, wepropose a new online framework enabling the use of ?1 sparse data fitting term in robust dictionary learning, notably enhancing the usability and practicality of this important technique. Extensive experiments have been carried out to validate our new framework.
2 0.95336157 392 cvpr-2013-Separable Dictionary Learning
Author: Simon Hawe, Matthias Seibert, Martin Kleinsteuber
Abstract: Many techniques in computer vision, machine learning, and statistics rely on the fact that a signal of interest admits a sparse representation over some dictionary. Dictionaries are either available analytically, or can be learned from a suitable training set. While analytic dictionaries permit to capture the global structure of a signal and allow a fast implementation, learned dictionaries often perform better in applications as they are more adapted to the considered class of signals. In imagery, unfortunately, the numerical burden for (i) learning a dictionary and for (ii) employing the dictionary for reconstruction tasks only allows to deal with relatively small image patches that only capture local image information. The approach presented in this paper aims at overcoming these drawbacks by allowing a separable structure on the dictionary throughout the learning process. On the one hand, this permits larger patch-sizes for the learning phase, on the other hand, the dictionary is applied efficiently in reconstruction tasks. The learning procedure is based on optimizing over a product of spheres which updates the dictionary as a whole, thus enforces basic dictionary proper- , ties such as mutual coherence explicitly during the learning procedure. In the special case where no separable structure is enforced, our method competes with state-of-the-art dictionary learning methods like K-SVD.
Author: Li He, Hairong Qi, Russell Zaretzki
Abstract: This paper addresses the problem of learning overcomplete dictionaries for the coupled feature spaces, where the learned dictionaries also reflect the relationship between the two spaces. A Bayesian method using a beta process prior is applied to learn the over-complete dictionaries. Compared to previous couple feature spaces dictionary learning algorithms, our algorithm not only provides dictionaries that customized to each feature space, but also adds more consistent and accurate mapping between the two feature spaces. This is due to the unique property of the beta process model that the sparse representation can be decomposed to values and dictionary atom indicators. The proposed algorithm is able to learn sparse representations that correspond to the same dictionary atoms with the same sparsity but different values in coupled feature spaces, thus bringing consistent and accurate mapping between coupled feature spaces. Another advantage of the proposed method is that the number of dictionary atoms and their relative importance may be inferred non-parametrically. We compare the proposed approach to several state-of-the-art dictionary learning methods super-resolution. tionaries learned resolution results ods. by applying this method to single image The experimental results show that dicby our method produces the best supercompared to other state-of-the-art meth-
4 0.92767727 66 cvpr-2013-Block and Group Regularized Sparse Modeling for Dictionary Learning
Author: Yu-Tseh Chi, Mohsen Ali, Ajit Rajwade, Jeffrey Ho
Abstract: This paper proposes a dictionary learning framework that combines the proposed block/group (BGSC) or reconstructed block/group (R-BGSC) sparse coding schemes with the novel Intra-block Coherence Suppression Dictionary Learning (ICS-DL) algorithm. An important and distinguishing feature of the proposed framework is that all dictionary blocks are trained simultaneously with respect to each data group while the intra-block coherence being explicitly minimized as an important objective. We provide both empirical evidence and heuristic support for this feature that can be considered as a direct consequence of incorporating both the group structure for the input data and the block structure for the dictionary in the learning process. The optimization problems for both the dictionary learning and sparse coding can be solved efficiently using block-gradient descent, and the details of the optimization algorithms are presented. We evaluate the proposed methods using well-known datasets, and favorable comparisons with state-of-the-art dictionary learning methods demonstrate the viability and validity of the proposed framework.
5 0.88366205 257 cvpr-2013-Learning Structured Low-Rank Representations for Image Classification
Author: Yangmuzi Zhang, Zhuolin Jiang, Larry S. Davis
Abstract: An approach to learn a structured low-rank representation for image classification is presented. We use a supervised learning method to construct a discriminative and reconstructive dictionary. By introducing an ideal regularization term, we perform low-rank matrix recovery for contaminated training data from all categories simultaneously without losing structural information. A discriminative low-rank representation for images with respect to the constructed dictionary is obtained. With semantic structure information and strong identification capability, this representation is good for classification tasks even using a simple linear multi-classifier. Experimental results demonstrate the effectiveness of our approach.
6 0.86421382 296 cvpr-2013-Multi-level Discriminative Dictionary Learning towards Hierarchical Visual Categorization
7 0.83736569 125 cvpr-2013-Dictionary Learning from Ambiguously Labeled Data
8 0.77842331 185 cvpr-2013-Generalized Domain-Adaptive Dictionaries
9 0.71291161 5 cvpr-2013-A Bayesian Approach to Multimodal Visual Dictionary Learning
10 0.64272809 422 cvpr-2013-Tag Taxonomy Aware Dictionary Learning for Region Tagging
11 0.60973167 220 cvpr-2013-In Defense of Sparsity Based Face Recognition
12 0.57776225 302 cvpr-2013-Multi-task Sparse Learning with Beta Process Prior for Action Recognition
13 0.5146327 204 cvpr-2013-Histograms of Sparse Codes for Object Detection
14 0.48134482 83 cvpr-2013-Classification of Tumor Histology via Morphometric Context
15 0.4561187 442 cvpr-2013-Transfer Sparse Coding for Robust Image Representation
16 0.41296992 399 cvpr-2013-Single-Sample Face Recognition with Image Corruption and Misalignment via Sparse Illumination Transfer
17 0.39910221 419 cvpr-2013-Subspace Interpolation via Dictionary Learning for Unsupervised Domain Adaptation
18 0.35675019 267 cvpr-2013-Least Soft-Threshold Squares Tracking
19 0.34178147 427 cvpr-2013-Texture Enhanced Image Denoising via Gradient Histogram Preservation
20 0.33407974 164 cvpr-2013-Fast Convolutional Sparse Coding
topicId topicWeight
[(10, 0.095), (12, 0.157), (15, 0.012), (16, 0.024), (26, 0.036), (28, 0.031), (33, 0.348), (67, 0.076), (69, 0.067), (87, 0.049)]
simIndex simValue paperId paperTitle
1 0.94824809 270 cvpr-2013-Local Fisher Discriminant Analysis for Pedestrian Re-identification
Author: Sateesh Pedagadi, James Orwell, Sergio Velastin, Boghos Boghossian
Abstract: Metric learning methods, , forperson re-identification, estimate a scaling for distances in a vector space that is optimized for picking out observations of the same individual. This paper presents a novel approach to the pedestrian re-identification problem that uses metric learning to improve the state-of-the-art performance on standard public datasets. Very high dimensional features are extracted from the source color image. A first processing stage performs unsupervised PCA dimensionality reduction, constrained to maintain the redundancy in color-space representation. A second stage further reduces the dimensionality, using a Local Fisher Discriminant Analysis defined by a training set. A regularization step is introduced to avoid singular matrices during this stage. The experiments conducted on three publicly available datasets confirm that the proposed method outperforms the state-of-the-art performance, including all other known metric learning methods. Furthermore, the method is an effective way to process observations comprising multiple shots, and is non-iterative: the computation times are relatively modest. Finally, a novel statistic is derived to characterize the Match Characteris- tic: the normalized entropy reduction can be used to define the ’Proportion of Uncertainty Removed’ (PUR). This measure is invariant to test set size and provides an intuitive indication of performance.
2 0.9363296 406 cvpr-2013-Spatial Inference Machines
Author: Roman Shapovalov, Dmitry Vetrov, Pushmeet Kohli
Abstract: This paper addresses the problem of semantic segmentation of 3D point clouds. We extend the inference machines framework of Ross et al. by adding spatial factors that model mid-range and long-range dependencies inherent in the data. The new model is able to account for semantic spatial context. During training, our method automatically isolates and retains factors modelling spatial dependencies between variables that are relevant for achieving higher prediction accuracy. We evaluate the proposed method by using it to predict 1 7-category semantic segmentations on sets of stitched Kinect scans. Experimental results show that the spatial dependencies learned by our method significantly improve the accuracy of segmentation. They also show that our method outperforms the existing segmentation technique of Koppula et al.
3 0.93240875 459 cvpr-2013-Watching Unlabeled Video Helps Learn New Human Actions from Very Few Labeled Snapshots
Author: Chao-Yeh Chen, Kristen Grauman
Abstract: We propose an approach to learn action categories from static images that leverages prior observations of generic human motion to augment its training process. Using unlabeled video containing various human activities, the system first learns how body pose tends to change locally in time. Then, given a small number of labeled static images, it uses that model to extrapolate beyond the given exemplars and generate “synthetic ” training examples—poses that could link the observed images and/or immediately precede or follow them in time. In this way, we expand the training set without requiring additional manually labeled examples. We explore both example-based and manifold-based methods to implement our idea. Applying our approach to recognize actions in both images and video, we show it enhances a state-of-the-art technique when very few labeled training examples are available.
same-paper 4 0.93190253 315 cvpr-2013-Online Robust Dictionary Learning
Author: Cewu Lu, Jiaping Shi, Jiaya Jia
Abstract: Online dictionary learning is particularly useful for processing large-scale and dynamic data in computer vision. It, however, faces the major difficulty to incorporate robust functions, rather than the square data fitting term, to handle outliers in training data. In thispaper, wepropose a new online framework enabling the use of ?1 sparse data fitting term in robust dictionary learning, notably enhancing the usability and practicality of this important technique. Extensive experiments have been carried out to validate our new framework.
5 0.92941386 170 cvpr-2013-Fast Rigid Motion Segmentation via Incrementally-Complex Local Models
Author: Fernando Flores-Mangas, Allan D. Jepson
Abstract: The problem of rigid motion segmentation of trajectory data under orthography has been long solved for nondegenerate motions in the absence of noise. But because real trajectory data often incorporates noise, outliers, motion degeneracies and motion dependencies, recently proposed motion segmentation methods resort to non-trivial representations to achieve state of the art segmentation accuracies, at the expense of a large computational cost. This paperproposes a method that dramatically reduces this cost (by two or three orders of magnitude) with minimal accuracy loss (from 98.8% achieved by the state of the art, to 96.2% achieved by our method on the standardHopkins 155 dataset). Computational efficiency comes from the use of a simple but powerful representation of motion that explicitly incorporates mechanisms to deal with noise, outliers and motion degeneracies. Subsets of motion models with the best balance between prediction accuracy and model complexity are chosen from a pool of candidates, which are then used for segmentation. 1. Rigid Motion Segmentation Rigid motion segmentation (MS) consists on separating regions, features, or trajectories from a video sequence into spatio-temporally coherent subsets that correspond to independent, rigidly-moving objects in the scene (Figure 1.b or 1.f). The problem currently receives renewed attention, partly because of the extensive amount of video sources and applications that benefit from MS to perform higher level computer vision tasks, but also because the state of the art is reaching functional maturity. Motion Segmentation methods are widely diverse, but most capture only a small subset of constraints or algebraic properties from those that govern the image formation process of moving objects and their corresponding trajectories, such as the rank limit theorem [9, 10], the linear independence constraint (between trajectories from independent motions) [2, 13], the epipolar constraint [7], and the reduced rank property [11, 15, 13]. Model-selection based (a)Orignalvideoframes(b)Clas -label dtrajectories (c)Modelsup ort egion(d)Modelinlersandcontrolpoints (e)Modelresiduals (f) Segmenta ion result Figure 1: Model instantiation and segmentation. a) fth original frame, Italian Grand Prix ?c 2012 Formula 1. b) Classilanbaell feradm, trajectory Gdaratan Wd P r(rixed ?,c green, bolrumeu alnad 1 .bbl a)c Ck correspond to chassis, helmet, background and outlier classes respectively). c) Spatially-local support subset for a candidate motion in blue. d) Candidate motion model inliers in red, control points from Eq. 3) in white. e) Residuals from Eq. 11) color-coded with label data, the radial coordinate is logarithmic. f) Segmentation result. Wˆf (rif (cif methods [11, 6, 8] balance model complexity with modeling accuracy and have been successful at incorporating more of these aspects into a single formulation. For instance, in [8] most model parameters are estimated automatically from the data, including the number of independent motions and their complexity, as well as the segmentation labels (including outliers). However, because of the large number of necessary motion hypotheses that need to be instantiated, as well as the varying and potentially very large number of 222222555977 model parameters that must be estimated, the flexibility offered by this method comes at a large computational cost. Current state of the art methods follow the trend of using sparse low-dimensional subspaces to represent trajectory data. This representation is then fed into a clustering algorithm to obtain a segmentation result. A prime example of this type of method is Sparse Subspace Clustering (SSC) [3] in which each trajectory is represented as a sparse linear combination of a few other basis trajectories. The assumption is that the basis trajectories must belong to the same rigid motion as the reconstructed trajectory (or else, the reconstruction would be impossible). When the assumption is true, the sparse mixing coefficients can be interpreted as the connectivity weights of a graph (or a similarity matrix), which is then (spectral) clustered to obtain a segmentation result. At the time of publication, SSC produced segmentation results three times more accurate than the best predecessor. The practical downside, however, is the inherently large computational cost of finding the optimal sparse representation, which is at least cubic on the number of trajectories. The work of [14] also falls within the class of subspace separation algorithms. Their approach is based on clustering the principal angles (CPA) of the local subspaces associated to each trajectory and its nearest neighbors. The clustering re-weights a traditional metric of subspace affinity between principal angles. Re-weighted affinities are then used for segmentation. The approach produces segmentation results with accuracies similar to those of SSC, but the computational cost is close to 10 times bigger than SSC’s. In this work we argue that competitive segmentation results are possible using a simple but powerful representation of motion that explicitly incorporates mechanisms to deal with noise, outliers and motion degeneracies. The proposed method is approximately 2 or 3 orders of magnitude faster than [3] and [14] respectively, currently considered the state of the art. 1.1. Affine Motion Projective Geometry is often used to model the image motion of trajectories from rigid objects between pairs of frames. However, alternative geometric relationships that facilitate parameter computation have also been proven useful for this purpose. For instance, in perspective projection, general image motion from rigid objects can be modeled via the composition of two elements: a 2D homography, and parallax residual displacements [5]. The homography describes the motion of an arbitrary plane, and the parallax residuals account for relative depths, that are unaccounted for by the planar surface model. Under orthography, in contrast, image motion of rigid objects can be modeled via the composition of a 2D affine transformation plus epipolar residual displacements. The 2D affine transformation models the motion of an arbitrary plane, and the epipolar residuals account for relative depths. Crucially, these two components can be computed separately and incrementally, which enables an explicit mechanism to deal with motion degeneracy. In the context of 3D motion, a motion is degenerate when the trajectories originate from a planar (or linear) object, or when neither the camera nor the imaged object exercise all of their degrees of freedom, such as when the object only translates, or when the camera only rotates. These are common situations in real world video sequences. The incremental nature of the decompositions described above, facilitate the transition between degenerate motions and nondegenerate ones. Planar Model Under orthography, the projection of trajectories from a planar surface can be modeled with the affine transformation: ⎣⎡xy1c ⎦⎤=?0D? 1t?⎡⎣yx1w ⎦⎤= A2wD→c⎣⎡yx1w ⎦⎤, (1) where D ∈ R2×2 is an invertible matrix, and t ∈ R2 is a threarnesl Datio ∈n v Rector. Trajectory icboloer mdiantartixe,s (axnwd , tyw ∈) are in the plane’s reference frame (modulo a 2D affine transformation) and (xc, yc) are image coordinates. Now, let W ∈ R2F×P be matrix of trajectory data that conNtaoiwns, tlehet x a n∈d y image coordinates of P feature points tracked through F frames, as in TocmputehW pa=r⎢m⎣ ⎡⎢ etyx e1F r.,s 1 ofA· . ·2.Dfyx r1Fo m., P tra⎦⎥ ⎤je.ctorydat,(l2e)t C = [c1, c2 , c3] ∈ R2f×3 be three columns (three full trajectories) from W, and let = be the x and y coordinates of the i-th control trajectory at frame f. Then the transformation between points from an arbitrary source frame s to a target frame f can be written as: cif [ci2f−1, ci2f]? ?c1f1 c12f c1f3?= A2sD→f?c11s and A2s→Df c12s c13s?, (3) ?−1. (4) can be simply computed as: A2sD→f= ? c11f c12f c13f ? ? c11s c12s c13s The inverse in the right-hand side matrix of Eq. 4 exists so long as the points cis are not collinear. For simplicity we refer to as and consequently A2sD is the identity matrix. A2s→Df A2fD 222222556088 3D Model In order to upgrade a planar (degenerate) model into a full 3D one, relative depth must be accounted using the epipolar residual displacements. This means extending Eq. 1 with a direction vector, scaled by the corresponding relative depth of each point, as in: ⎣⎡xy1c ⎦⎤=?0?D? t1? ⎡⎣xy1w ⎦⎤+ δzw⎣⎡a 0213 ⎦⎤. The depth δzw is relative to the arbitrary plane tion is modeled by A2D; a point that lies on would have δzw = 0. We call the orthographic the plane plus parallax decomposition, the 2D Epipolar (2DAPE) decomposition. Eq. 5 is equivalent to (5) whose mosuch plane version of Affine Plus wher⎣⎡ ixyt1c is⎤⎦cl=ear⎡⎣tha 120t hea p201a2rma2e10t3erst1o2f⎦⎤A⎣⎢⎡3Dδyxd1zwefin⎦⎥⎤ean(o6r)thographically projected 3D affine transformation. Deter- mining the motion and structure parameters of a 3D model from point correspondences can be done using the classical matrix factorization approach [10], but besides being sensitive to noise and outliers, the common scenario where the solution becomes degenerate makes the approach difficult to use in real-world applications. Appropriately accommodating and dealing with the degenerate cases is one of the key features of our work. 2. Overview of the Method The proposed motion segmentation algorithm has three stages. First, a pool of M motion model hypotheses M = s{tMag1e , . . . , rMst,M a} p oiso generated using a omdeetlh hoydp tohthate csoesm Mbine =s a MRandom Sampling naenrda eCdon usseinngsu as m(ReAthNodS tAhaCt) [o4m] bteinche-s nique with the 2DAPE decomposition. The goal is to generate one motion model for each of the N independent, rigidly-moving objects in the scene; N is assumed to be known a priori. The method instantiates many more models than those expected necessary (M ? N) in an attempt ienlscr tehaasne tthhoes elik eexlpiheocotedd o nfe generating Mco ?rrec Nt m) iond aenl proposals for all N motions. A good model accurately describes a large subset of coherently moving trajectories with the smallest number of parameters (§3). Ialnl ethste n suemcobnedr stage, msubetseertss o§f3 )m.otion models from M are ncom thebi sneecdo ntod explain ualbl sthetes trajectories mino tdheel sequence. The problem is framed as an objective function that must be minimized. The objective function is the negative loglikelihood over prediction accuracy, regularized by model complexity (number of model parameters) and modeling overlap (trajectories explained by multiple models). Notice that after this stage, the segmentation that results from the optimal model combination could be reported as a segmentation result (§5). ioTnhe r tshuilrtd ( stage incorporates the results from a set of model combinations that are closest to the optimal. Segmentation results are aggregated into an affinity matrix, which is then passed to a spectral clustering algorithm to produce the final segmentation result. This refinement stage generally results in improved accuracy and reduced segmentation variability (§6). 3. Motion Model Instantiation Each model M ∈ M is instantiated independently using RacAhN mSAodCel. MThis ∈ c Mhoi cies niss manotitiavteatded in bdeecpaeunsdee otlfy th usismethod’s well-known computational efficiency and robustness to outliers, but also because of its ability to incorporate spatially local constraints and (as explained below) because most of the computations necessary to evaluate a planar model can be reused to estimate the likelihoods of a potentially necessary 3D model, yielding significant computational savings. The input to our model instantiation algorithm is a spatially-local, randomly drawn subset of trajectory data Wˆ[2F×I] ⊆ W[2F×P] (§3.1). In turn, at each RANSAC trial, the algorithm draw(§s3 uniformly d,i asttr eibaucthed R, A rNanSdoAmC subsets of three control trajectories (C[2F×3] ⊂ Wˆ[2F×I]). Each set of control trajectories is used to estim⊂ate the family of 2D affine transformations {A1, . . . , AF} between the iblyase o ffr 2aDm aef ainnde aralln sotfoherrm fartaimoness { iAn the sequence, wtwheicehn are then used to determine a complete set of model parameters M = {B, σ, C, ω}. The matrix B ∈ {0, 1}[F×I] indicates Mwhe =the {rB t,hσe ,iC-th, trajectory asthroixu Bld ∈b e predicted by model M at frame f (inlier, bif = 1) or not (outlier, bif = 0), σ = {σ1 , . . . , σF} are estimates of the magnitude of the σnois =e {foσr each fram}e a, aen eds ω ∈at {s2 oDf, t3hDe} m isa tnhietu edsetim ofa ttehde nmooidseel f type. hTh fera goal aisn dto ω ωfin ∈d {t2heD c,3oDntr}ol is points tainmda ttehed associated parameters that minimize the objective function O(Wˆ,M) =f?∈Fi?∈IbifLω? wˆif| Af,σf?+ Ψ(ω) + Γ(B) across (7) wˆfi a number of RANSAC trials, where = = are the coordinates of the i-th trajectory from the support subset at frame f. The negative log-likelihood term Lω (·) penalizes reconstruction error, while Ψ(·) and Γ(·) are regularizers. Tcohen tthrureceti otenr mer-s are ,d wefhinileed Ψ Ψ b(e·l)ow an. Knowing that 2D and 3D affine models have 6 and 8 degrees of freedom respectively, Ψ(ω) regularizes over model complexity using: (xif, yif) ( wˆ 2if−1, wˆ i2f) Wˆ Ψ(ω) =?86((FF − − 1 1)), i f ωω== 32DD. 222222556199 (8) Γ(B) strongly penalizes models that describe too few trajectories: Γ(B) =?0∞,, oifth?erwI?iseFbif< Fλi (9) The control set C whose M minimizes Eq. 7 across a number of RANSAC trials becomes part of the pool of candidates M. 2D likelihoods. For the planar case (ω = 2D) the negative log-likelihood term is evaluated with: L2D( wˆif| Af,σf) = −log?2π|Σ1|21exp?−21rif?Σ−1rif??, (10) which is a zero-mean 2D Normal distribution evaluated at the residuals The spherical covariance matrix is Σ = rif. rif (σf)2I. The residuals are determined by the differences between the predictions made by a hypothesized model Af, and the observations at each frame ?r?1f?=? w˜1?f?− Af? w˜1?s?. (11) 3D likelihoods. The negative log-likelihood term for the 3D case is based on the the 2DAPE decomposition. The 2D affinities Af and residuals rf are reused, but to account for the effect of relative depth, an epipolar line segment ef is robustly fit to the residual data at each frame (please see supplementary material for details on the segment fitting algorithm). The 2DAPE does not constrain relative depths to remain constant across frames, but only requires trajectories to be close to the epipolar line. So, if the unitary vector ef indicates the orthogonal direction to ef, then the negativ⊥e log-likelihood term for the 3D case is estimated with: L3D( wˆfi| Af,σf) = −2log⎜⎝⎛√21πσfexp⎪⎨⎪⎧−?r2if(?σfe)f⊥2?2⎪⎬⎪⎫⎞⎟⎠, ⎠(12,) which is also a zero-mean 2D Norma⎩l distribution ⎭computed as the product of two identical, separable, singlevariate, normal distributions, evaluated at the distance from the residual to the epipolar line. The first one corresponds to the actual deviation in the direction of ef , which is analyti- cally computed using rif?ef. The seco⊥nd one corresponds to an estimate of the deviat⊥ion in the perpendicular direction (ef), which cannot be determined using the 2DAPE decomposition model, but can be approximated to be equal to rif ? ef, which is a plausible estimate under the isotropic noise as⊥sumption. Note that Eq. 7 does not evaluate the quality of a model using the number of inliers, as it is typical for RANSAC. Instead, we found that better motion models resulted from Algorithm 1: Motion model instantiation × Algorithm 1: Motion model instantiation Input:b Traasejec frtoamrye d bata W[2F×P], number of RANSAC trials K, arbitrary Output: Parameters of the motion model M = {B , σn , ω} // determine the training set c ← rand( 1, P) ; r ← rand(rmin , rmax ) // random center and radius I] ← t ra j e ct oriesWithinDis k (W, r,c) // support subset X ← homoCoords(Wˆb) // points at base frame for K RANSAC trials do Wˆ[2F Wˆ return M = {B} optimizing over the accuracy ofthe model predictions for an (estimated) inlier subset, which also means that the effect of outliers is explicitly uncounted. Figure 1.b shows an example of class-labeled trajectory data, 1.c shows a typical spatially-local support subset. Figures 1.d and 1.e show a model’s control points and its corresponding (class-labeled) residuals, respectively. A pseudocode description of the motion instantiation algorithm is provided in Algorithm 1. Details on how to determine Wˆ, as well as B, σ, and ω follow. 3.1. Local Coherence The subset of trajectories Wˆ given to RANSAC to generate a model M is constrained to a spatially local region. The probability ofchoosing an uncontaminated set of 3 control trajectories, necessary to compute a 2D affine model, from a dataset with a ratio r of inliers, after k trials is: p = 1 − (1 − r3)k. This means that the number of trials pne =ede 1d −to (fi1n d− a subset of 3 inliers with probability p is k =lloogg((11 − − r p3)). (13) A common assumption is that trajectories from the same underlying motion are locally coherent. Hence, a compact region is likely to increase r, exponentially reducing 222222666200 Figure 2: Predictions (red) from a 2D affine model with standard Gaussian noise (green) on one of the control points (black). Noiseless model predictions in blue. All four scenarios have identical noise. The magnitude of the extrapolation error changes with the distance between the control points. k, and with it, RANSAC’s computation time by a proportional amount. The trade-off that results from drawing model control points from a small region, however, is extrapolation error. A motion model is extrapolated when utilized to make predictions for trajectories outside the region defined by the control points. The magnitude of modeling error depends on the magnitude of the noise affecting the control points, and although hard to characterize in general, extrapolation error can be expected to grow with the distance from the prediction to the control points, and inversely with the distance between the control points themselves. Figure 2 shows a series of synthetic scenarios where one of the control points is affected by zero mean Gaussian noise of small magnitude. Identical noise is added to the same trajectory in all four scenarios. The figure illustrates the relation between the distance between the control points and the magnitude of the extrapolation errors. Our goal is to maximize the region size while limiting the number of outliers. Without any prior knowledge regarding the scale of the objects in the scene, determining a fixed size for the support region is unlikely to work in general. Instead, the issue is avoided by randomly sampling disk-shaped regions of varying sizes and locations to construct a diverse set of support subsets. Each support subset is then determined by Wˆ = {wi | (xbi − ox)2 + (ybi − oy)2 < r2}, (14) where (ox , oy) are the coordinates of the center of a disk of radius r. To promote uniform image coverage, the disk is centered at a randomly chosen trajectory (ox , oy) = (xbi, yib) with uniformly distributed i ∼ U(1, P) and base frame b) w∼i h U u(1n,i fFor)m. yTo d asltrloibwu efodr idi ∼ffer Ue(n1t, region ds bizaesse, tfhraem read bius ∼ r is( ,cFho)s.en T ofro amllo a u fnoirfo dirmffe rdeinsttr riebugtiioonn r ∼s, tUh(erm raidni,u ursm rax i)s. Ihfo tsheenre f are mI a trajectories swtritihbiunt othne support region, then ∈ R2F×I. It is worth noting that the construction of theW support region does not incorporate any knowledge about the motion of objects in the scene, and in consequence will likely contain trajectories that originate from more than one independently moving object (Figure 3). Wˆ Wˆ Figure 3: Two randomly drawn local support sets. Left: A mixed set with some trajectories from the blue and green classes. Right: Another mixed set with all of the trajectories in the red class and some from the blue class. 4. Characterizing the Residual Distribution At each RANSAC iteration, residuals rf are computed using the 2D affine model Af that results from the constraints provided by the control trajectories C. Characterizing the distribution of rf has three initial goals. The first one is to determine 2D model inliers b2fD (§4.1), the second one is to compute estimates of the magnitude ,o tfh thee s ncooinsed at every frame σ2fD (§4.2), and the third one is to determine whether the residual( §d4i.s2t)r,ib auntidon th originates efr iosm to a planar or a 3D object (§4.3). If the object is suspected 3D, then two more goals n (§e4ed.3 )to. bIfe t achieved. sT shues pfiercstt one Dis, t hoe nde ttweromine 3D model inliers b3fD (§4.4), and the second one is to estimate the magnitude of the noise of a 3D model (§4.5). (σf3D) to reflect the use 4.1. 2D Inlier Detection Suppose the matrix Wˆ contains trajectories Wˆ1 ∈ R2F×I and Wˆ2 ∈ R2F×J from two independently moving objects, and ∈tha Rt these trajectories are contaminated with zero-mean Gaussian noise of spherical covariance η ∼ N(0, (σf)2I): Wˆ = ?Wˆ1|Wˆ2? + η. (15) A1f Now, assume we know the true affine transformations and that describe the motion of trajectories for the subsets Wˆ1 and Wˆ2, respectively. If is used to compute predictions for all of Wˆ (at frame f), the expected value (denoted by ?·?) of the magnitude of the residuals (rf from Eq. 11) for trajectories aing nWiˆtud1 will be in the order of the magnitude of the underlying noise ?|rif |? = σf for each i∈ {1, . . . , I}. But in this scenario, trajectories in Wˆ2 ewaicl h b ie ∈ predicted using tth ien wrong emnaodrioel,, resulting isn i nr esid?uals? wit?h magnitudes de?termined by the motion differential A2f ???rif??? A1f ???(A1f − A2f) wˆib???. If we = can assume that the motion ?d??riff???er =en???t(iAal is bigger tha???n. tIhfe w deis cpalnac aesmsuemnte d thuea t toh eno miseo:t ???(A1f − A2f)wib??? 222222666311 > σf, (16) then the model inliers can be determined by thresholding | with the magnitude of the noise, scaled by a constant |(τr =| w wλitσhσtf h):e |rif bif=?10,, |orthife|r ≤wi τse. (17) But because σf is generally unknown, the threshold (τ) is estimated from the residual data. To do so, let be the vector of residual magnitudes where rˆi ≤ ˆ ri+1 . Now, let r˜ = median ( rˆi+1 −ˆ r i). The threshold i≤s trˆ h en defined as r τ = min{ rˆi | (ri+1 − ri) > λr˜ r}, (18) which corresponds to the smallest residual magnitude before a salient magnitude gap. Our experiments showed this test to be efficient and effective. Figure 1.e shows classlabeled residuals. Notice the presence of a (low density) gap between the residuals from the trajectories explained by the correct model (in red, close to the origin), and the rest. 4.2. Magnitude of the Noise, 2D Model r2fD Let contain only the residuals of the inlier trajectories (those where = 1), and let USV? be the singular value decomposition of the covariance matrix bif ofˆ r2fD: USV?= svd??1bpf( rˆ2fD)? rˆ2fD?.(19) Then the magnitude of the n?oise corresponds to the largest singular value σ2 = s1, because if the underlying geometry is in fact planar, then the only unaccounted displacements captured by the residuals are due to noise. Model capacity can also be determined from S, as explained next. 4.3. Model Capacity The ratio of largest over smallest singular values (s1/s2) determines when upgrading to a 3D model is beneficial. When the underlying geometry is actually non-planar, the residuals from a planar model should distribute along a line (the epipolar line), reflecting that their relative depth is being unaccounted for. This produces a covariance matrix with a large ratio s1/s2 ? 1. If on the other hand, if s1/s2 ≈ 1, then there is no in 1d.ic Iafti oonn tohfe unexplained relative depth, tihn wnh thicehr case, fitting a olinne o tfo u spherically distributed residuals will only increase the model complexity without explaining the residual variance much better. A small spherical residual covariance strongly suggests a planar underlying geometry. 4.4. 3D Inlier Detection When the residual distribution is elongated (s1/s2 ? 1), a line segment is robustly fit to the (potentially con?tam 1i)-, nated) set of residuals. The segment must go through the origin and its parameters are computed using a Hough transform. Further details about this algorithm can be found in the supplementary material. Inlier detection The resulting line segment is used to determine 3D model inliers. Trajectory ibecomes an inlier at frame f if it satisfies two conditions. First, the projection of rif onto the line must lie within the segment limits (β ≤ ≤ γ). Second, the normalized distance to the rif?ef (ef?rif line must be below a threshold ≤ σ2λd). Notice that the threshold depends on the smalle≤st singular value from Eq. 19 to (roughly) account for the presence of noise in the direction perpendicular to the epipolar (ef). 4.5. Magnitude of the Noise, 3D Model let rˆf3D Similarly to the 2D case, contain the residual data from the corresponding 3D inlier trajectories. An estimate for the magnitude of the noise that reflects the use of a 3D model can be obtained from the singular value decomposition of the covariance matrix of r3fD (as in Eq. 19). In this case, the largest singular value s1 captures the spread of residuals along the epipolar line, so its magnitude is mainly related to the magnitude of the displacements due to relative depth. However, s2 captures deviations from the epipolar line, which in a rigid 3D object can only be attributed to noise, making σ2 = s2 a reasonable estimate for its magnitude. Optimal model parameters When both 2D and 3D models are instantiated, the one with the smallest penalized negative log-likelihood (7) becomes the winning model for the current RANSAC run. The same penalized negative loglikelihood metric is used to determine the better model from across all RANSAC iterations. The winning model is added to the pool M, and the process is repeated M times, forming hthee p pool MM, a=n d{ tMhe1 , . . . , MssM is} r.e 5. Optimal Model Subset The next step is to find the model combination M? ⊂ M thhea tn mxta xstiempiz iess t prediction accuracy finora othne Mwhol⊂e trajectory mdaaxtaim Wize,s w phreiledi minimizing cmyod foerl complexity and modelling overlap. For this purpose, let Mj = {Mj,1 , . . . , Mj,N} be the j-th m thoisdel p ucorpmosbein,at lieotn, M Mand let {Mj} be the set o}f baell MheC jN-th = m N!(MM−!N)!) caotimonb,in aantdio lnest of N-sized models than can be draNw!(nM fr−oNm) M). The model soefle Nct-sioinze problem sis t hthanen c faonr bmeu dlartaewdn as M?= ar{gMmj}inOS(Mj), (20) 222222666422 where the objective is ?N ?P OS(Mj) = ??πp,nE (wp,Mj,n) ?n=1p?P=1 + λΦ?Φ(wp,Mj,n) + λΨ?Ψ(Mj,n). ?N (21) i?= ?1 n?= ?1 The first term accounts for prediction accuracy, the other two are regularization terms. Details follow. Prediction Accuracy In order to determine how well a model M predicts an arbitrary trajectory w, the affine transformations estimated by RANSAC could be re-used. However, the inherent noise in the control points, and the potentially short distance between them, often render this approach impractical, particularly when w is spatially distant from the control points (see §3. 1). Instead, model parametferorsm are computed owinittsh a efeac §to3r.1iz)a.ti Ionnst e baadse,d m [o1d0e]l mpaertahmode.Given the inlier labeling B in M, let WB be the subset of trajectories where bif = 1for at least half of the frames. The orthonormal basis S of a ω = 2D (or 3D) motion model can be determined by the 2 (or 3) left singular vectors of WB. Using S as the model’s motion matrices, prediction accuracy can be computed using: E(w, M) = ??SS?w − w??2 , (22) which is the sum of squared?? Euclidean d??eviations from the predictions (SS?w), to the observed data (w). Our experiments indicated that, although sensitive to outliers, these model predictions are much more robust to noise. Ownership variables Π ∈ {0, 1}[P×N] indicate whether trajectory p i ps explained by t {he0 ,n1-}th model (πp,n = 1) or not (πp,n = 0), and are determined by maximum prediction accuracy (i.e. minimum Euclidean deviation): πp,n=⎨⎧01,, oift hMerjw,nis=e. aMrg∈mMinjE(wp,M) (23) Regularization terms The second term from Eq. 21 penalizes situations where multiple models explain a trajectory (w) with relatively small residuals. For brevity, let M) = exp{−E(w, M)}, then: Eˆ(w, Φ(w,Mj) = −logMMm∈?∈aMMxjE ˆ ( w , MM)). (24) The third term regularizes over the number of model parameters, and is evaluated using Eq. 8. The constants λΦ and λΨ modulate the effect of the corresponding regularizer. Table 1: Accuracy and run-time for the H155 dataset. Naive RANSAC included as a baseline with overall accuracy and total computation time estimated using data from [12]. SOCARAPSulgrCAoNs[S r[31itA]4h]CmAverage89 A689 c. 71c 7695u racy[%]Compu1 t4a 237t1i506o70 n0 time[s] 6. Refinement The optimal model subset M? yields ownership variableTsh Πe o? wtimhicahl can already tb e M interpreted as a segmentation result. However, we found that segmentation accuracy can be improved by incorporating the labellings Πt from the top T subsets {Mt? | 1 ≤ t ≤ T} closest to optimal. Multiple labellings are incorporated oinsetos an affinity matrix F, where the fi,j entry indicates the frequency with which trajectory i is given the same label as trajectory j across all T labelli?ngs, weighted b?y the relative objective function O˜t = exp ?−OOSS((WW||MMt??))? for such a labelling: fi,j=?tT=11O˜tt?T=1?πit,:πjt,?:?O˜t (25) Note that the inne?r product between the label vectors (πi,:πj?,:) is equal to one only when the labels are the same. A spectral clustering method is applied on F to produce the method’s final segmentation result. 7. Experiments Evaluation was made through three experimental setups. Hopkins 155 The Hopkins 155 (H155) dataset has been the standard evaluation metric for the problem of motion segmentation of trajectory data since 2007. It consists of checkerboard, traffic and articulated sequences with either 2 or 3 motions. Data was automatically tracked, but tracking errors were manually corrected; further details are available in [12]. The use of a standard dataset enables direct comparison of accuracy and run-time performance. Table 1 shows the relevant figures for the two most competitive algorithms that we are aware of. The data indicates that our algorithm has run-times that are close to 2 or 3 orders of magnitude faster than the state of the art methods, with minimal accuracy loss. Computation times are measured in the same (or very similar) hardware architectures. Like in CPA, our implementation uses a single set of parameters for all the experiments, but as others had pointed out [14], it remains unclear whether the same is true for the results reported in the original SSC paper. 222222666533 Figure 4: Accuracy error-bars across artificial H155 datasets with controlled levels of Gaussian noise. Artificial Noise The second experimental setup complements an unexplored dimension in the H155 dataset: noise. The goal is to determine the effects of noise of different magnitudes towards the segmentation accuracy of our method, in comparison with the state of the art. We noted that H155 contains structured long-tailed noise, but for the purpose of this experiment we required a noise-free dataset as a baseline. To generate such a dataset, ground-truth labels were used to compute a rank 3 reconstruction of (mean-subtracted) trajectories for each segment. Then, multiple versions of H155 were computed by contaminating the noise-free dataset with Gaussian noise of magnitudes σn ∈ {0.01, 0.25, 0.5, 1, 2, 4, 8}. Our method, as well as SSC a∈nd { 0C.0PA1, were run on t2h,e4s,e8 }n.oi Ose-ucro mnterothlloedd, datasets; results are shown in Figure 4. The error bars on SSC and Ours indicate one standard deviation, computed over 20 runs. The plot for CPA is generated with only one run for each dataset (running time: 11.95 days). The graph indicates that our method only compromises accuracy for large levels of noise, while still being around 2 or 3 orders of magnitude faster than the most competitive algorithms. KLT Tracking The last experimental setup evaluates the applicability of the algorithm in real world conditions using raw tracks from an off-the-shelf implementation [1] of the Kanade-Lucas-Tomasi algorithm. Several sequences were tracked and the resulting trajectories classified by our method. Figure 5 shows qualitatively good motion segmentation results for four sequences. Challenges include very small relative motions, tracking noise, and a large presence of outliers. 8. Conclusions We introduced a computationally efficient motion segmentation algorithm for trajectory data. Efficiency comes from the use of a simple but powerful representation of motion that explicitly incorporates mechanisms to deal with noise, outliers and motion degeneracies. Run-time comparisons indicate that our method is 2 or 3 orders of magnitude faster than the state of the art, with only a small loss in accuracy. The robustness of our method to Gaussian noise tracks from four Formula 1 sequences. Italian Grand Prix ?c2012 Formula 1. In this figure, all trajectories are given a m?co2ti0o1n2 l Faoberml, including hoiust fliigeurrse. of different magnitudes was found competitive with state of the art, while retaining the inherent computational efficiency. The method was also found to be useful for motion segmentation of real-world, raw trajectory data. References [1] ht tp : / /www . ce s . c l emn s on . edu / ˜stb / k lt . 8 [2] J. P. Costeira and T. Kanade. A Multibody Factorization Method for Independently Moving Objects. IJCV, 1998. 1 [3] E. Elhamifar and R. Vidal. Sparse subspace clustering. In Proc. CVPR, 2009. 2, 7 [4] M. A. Fischler and R. C. Bolles. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM, 1981. 3 [5] M. Irani and P. Anandan. Parallax geometry of pairs of points for 3d scene analysis. Proc. ECCV, 1996. 2 [6] K. Kanatani. Motion segmentation by subspace separation: Model selection and reliability evaluation. International Journal Image Graphics, 2002. 1 [7] H. Longuet-Higgins. A computer algorithm for reconstructing a scene from two projections. Readings in Computer Vision: Issues, Problems, Principles, and Paradigms, MA Fischler and O. Firschein, eds, 1987. 1 [8] K. Schindler, D. Suter, , and H. Wang. A model-selection framework for multibody structure-and-motion of image sequences. Proc. IJCV, 79(2): 159–177, 2008. 1 [9] C. Tomasi and T. Kanade. Shape and motion without depth. Proc. [10] [11] [12] [13] [14] [15] ICCV, 1990. 1 C. Tomasi and T. Kanade. Shape and motion from image streams under orthography: a factorization method. IJCV, 1992. 1, 3, 7 P. Torr. Geometric motion segmentation and model selection. Phil. Tran. of the Royal Soc. of Lon. Series A: Mathematical, Physical and Engineering Sciences, 1998. 1 R. Tron and R. Vidal. A Benchmark for the Comparison of 3-D Motion Segmentation Algorithms. In Proc. CVPR, 2007. 7 J. Yan and M. Pollefeys. A factorization-based approach for articulated nonrigid shape, motion and kinematic chain recovery from video. PAMI, 2008. 1 L. Zappella, E. Provenzi, X. Llad o´, and J. Salvi. Adaptive motion segmentation algorithm based on the principal angles configuration. Proc. ACCV, 2011. 2, 7 L. Zelnik-Manor and M. Irani. Degeneracies, dependencies and their implications in multi-body and multi-sequence factorizations. Proc. CVPR, 2003. 1 222222666644
6 0.9181459 94 cvpr-2013-Context-Aware Modeling and Recognition of Activities in Video
7 0.91735119 82 cvpr-2013-Class Generative Models Based on Feature Regression for Pose Estimation of Object Categories
8 0.91730404 36 cvpr-2013-Adding Unlabeled Samples to Categories by Learned Attributes
9 0.91677725 4 cvpr-2013-3D Visual Proxemics: Recognizing Human Interactions in 3D from a Single Image
10 0.91674554 355 cvpr-2013-Representing Videos Using Mid-level Discriminative Patches
11 0.91655064 318 cvpr-2013-Optimized Pedestrian Detection for Multiple and Occluded People
12 0.91623509 221 cvpr-2013-Incorporating Structural Alternatives and Sharing into Hierarchy for Multiclass Object Recognition and Detection
13 0.91603386 446 cvpr-2013-Understanding Indoor Scenes Using 3D Geometric Phrases
14 0.91578829 43 cvpr-2013-Analyzing Semantic Segmentation Using Hybrid Human-Machine CRFs
15 0.9157806 204 cvpr-2013-Histograms of Sparse Codes for Object Detection
16 0.91545522 416 cvpr-2013-Studying Relationships between Human Gaze, Description, and Computer Vision
17 0.91534454 284 cvpr-2013-Mesh Based Semantic Modelling for Indoor and Outdoor Scenes
18 0.91498154 70 cvpr-2013-Bottom-Up Segmentation for Top-Down Detection
19 0.91497976 67 cvpr-2013-Blocks That Shout: Distinctive Parts for Scene Classification
20 0.91452044 438 cvpr-2013-Towards Pose Robust Face Recognition