cvpr cvpr2013 cvpr2013-105 knowledge-graph by maker-knowledge-mining
Source: pdf
Author: Fei Chen, Huimin Yu, Roland Hu, Xunxun Zeng
Abstract: In this paper we introduce a new shape-driven approach for object segmentation. Given a training set of shapes, we first use deep Boltzmann machine to learn the hierarchical architecture of shape priors. This learned hierarchical architecture is then used to model shape variations of global and local structures in an energetic form. Finally, it is applied to data-driven variational methods to perform object extraction of corrupted data based on shape probabilistic representation. Experiments demonstrate that our model can be applied to dataset of arbitrary prior shapes, and can cope with image noise and clutter, as well as partial occlusions.
Reference: text
sentIndex sentText sentNum sentScore
1 Given a training set of shapes, we first use deep Boltzmann machine to learn the hierarchical architecture of shape priors. [sent-2, score-0.614]
2 This learned hierarchical architecture is then used to model shape variations of global and local structures in an energetic form. [sent-3, score-0.453]
3 Finally, it is applied to data-driven variational methods to perform object extraction of corrupted data based on shape probabilistic representation. [sent-4, score-0.266]
4 Experiments demonstrate that our model can be applied to dataset of arbitrary prior shapes, and can cope with image noise and clutter, as well as partial occlusions. [sent-5, score-0.12]
5 Without utilizing any high-level prior information about expected objects, purely low-level information such as intensity, color and texture does not provide the desired segmentations. [sent-8, score-0.071]
6 In numerous studies [1-6], prior knowledge about the shapes of the objects to be segmented can significantly improve the final reliability and accuracy of the segmentation result. [sent-9, score-0.247]
7 However, given a training set of arbitrary prior shapes, there remains an open problem of how to define an appropriate prior shape model to guide object segmentation. [sent-10, score-0.416]
8 The shape of an object is represented as a set of points. [sent-14, score-0.172]
9 The evolutional shape is constrained by the point distribution model which is inferred from a training set of shapes. [sent-16, score-0.361]
10 Later, level set based approaches have gained significant attention toward the integration of shape prior into variational segmentation [2-7]. [sent-18, score-0.337]
11 Almost all these works optimize a linear combination of a data-driven term and a shape constraint term. [sent-19, score-0.172]
12 Data-driven term aims at driving the segmenting curve to the object boundaries, and shape constraint term restricts possible shapes embodied by the contour. [sent-20, score-0.447]
13 In level set approaches, shape is represented implicitly by signed distance functions (SDF). [sent-21, score-0.172]
14 This shape representation is consistent with the level set framework, and has its advantages since parameterization free and easy handling of topological changes. [sent-22, score-0.172]
15 However, SDF for shape representtation are not closed under linear operations, e. [sent-23, score-0.172]
16 , the mean shape and linear combinations of training shapes are typically no longer SDF. [sent-25, score-0.4]
17 Most existing works only consider rseimprielsaer nptraitioorn s hoaf spheas eof d ae iknneodw ans ao mbjaepcpt cnlgas ? [sent-26, score-0.242]
18 trreehaparts eosanes,s gCantirsoe nmt ooe rfse s areypt apdile. [sent-37, score-0.061]
19 trehpart esaessnigtantsio nto oo fe svhearyp e pdixefeiln ? [sent-50, score-0.121]
20 rptheropatbr easabesisnli gtatnyt ot hnto aot e tvshheisra ep ip dxixeeefli ni s? [sent-74, score-0.1]
21 tehex data-driven function optimized on convex shape spaces of the? [sent-91, score-0.172]
22 inrsde p, ar reessshepnaeptc etth iveoe frl eypg, riowobnha bldeiel stischteri cpl taorsertps t oeersf m met hn? [sent-257, score-0.064]
23 inrsed p,a r eessshepnaetpc ett hiveoe fr leypg, iwoobhnai ldbeei stichsrtei pl atorsetrsp t reoersfm ethn ? [sent-268, score-0.163]
24 e Tckhgerroeu anrde, m reasnpye wctiavyesl yt,o wdehfiilnee thhee lsahsatp tee rcmon ? [sent-276, score-0.056]
25 Simple uniform distribution [4], Gaussian densities [2], non-parametric estimator [5, 6], manifold learning [9, 10], and sparse representation [11] were considered to model shape variation within a training set. [sent-278, score-0.224]
26 They are suitable for segmenting objects of a known class in the image according to their possible similar shapes. [sent-280, score-0.049]
27 If the given training set of shapes is large and associated with multiple different object classes, the statistical shape models and manifold learning do not effectively represent the shape distributions due to large variability of shapes. [sent-281, score-0.572]
28 In addition, global transformations like translation, rotation and scaling and local transformations like bending and stretching are expensive to shape model in image 111888667088 Table 1. [sent-282, score-0.228]
29 Recently, deep learning models [12, 13] are attractive for their well performance in modeling high-dimensional richly structured data. [sent-288, score-0.302]
30 A deep learning model is about learning multiple levels of representation and abstraction that help to make sense of data such as images, sound, and text. [sent-289, score-0.246]
31 The deep Boltzmann machine (DBM) has been an important development in the quest for powerful deep learning models [14, 15]. [sent-290, score-0.553]
32 Applications that use deep learning models are numerous in computer vision and information retrieval, including classification [15], dimensionality reduction [12], visual recognition tasks [16], acoustic modeling [17], etc. [sent-291, score-0.299]
33 This shape generative model has the appealing property that it can both generate realistic samples and generalize to generate samples that differ from shapes in the training set. [sent-293, score-0.4]
34 Inspired by the above work [18], we focus on image segmentation, and propose a shape prior constraint term by deep learning to guide variational segmentation. [sent-294, score-0.633]
35 In this paper, we first use deep Boltzmann machine to extract the hierarchical architecture of shapes in the training set. [sent-295, score-0.618]
36 This architecture can effectively capture global and local features of prior shapes. [sent-296, score-0.215]
37 It is then introduced into the variational framework as a shape prior term in an energetic form. [sent-297, score-0.474]
38 By minimizing the proposed objective functional, the model is able to constrain an evolutional shape to follow global shape consistency while preserving its ability to capture local deformations. [sent-298, score-0.481]
39 Learning shape priors via DBM A Restricted Boltzmann Machine (RBM) is a particular type of Markov Random Field (MRF) that has a two-layer architecture, in which the visible units are connected to hidden units. [sent-300, score-0.394]
40 A Deep Boltzmann Machine (DBM) is an extension of the RBM that has multiple layers of hidden units arranged in layers [14]. [sent-301, score-0.137]
41 In general, the shape prior can be simply described as two levels of representation: low-level local features (like edges or corners) and high-level global features (like object parts or object). [sent-302, score-0.243]
42 On the other hand, high-level global features describe the image content, and they are more appropriate to cope with occlusion, noise, and (a) (b) Fig. [sent-304, score-0.049]
43 (a) The visible-to-hidden weights receive inputs only from a square patch of visible units below. [sent-307, score-0.415]
44 (b) A simple case that the training shape is divided into four square patches. [sent-308, score-0.29]
45 th hiag" ht ierse rd p pberfei nsnaeernydt hthiedL dsehetna ? [sent-341, score-0.059]
46 nt sri is(c a lsinot ekrnaocwtionn a st ebrimasse, s*),? [sent-462, score-0.133]
47 aanndd +* i s the visible ssvheeiidsllfifdb--ecclneoo n nvnsneyeecmcctot imiroo e? [sent-463, score-0.085]
48 r iT (sc ah lesin opte rkronabcoatwiboinln ai t sye brtihmaastse ts*h),e amannodd? [sent-465, score-0.166]
49 7", the leaGrnivinegn ao sf e D ofB Mali ncoends tirsatisn nogf hdeapteersm ! [sent-588, score-0.136]
50 r7e"l,a ttehde weGigivhtesn aan dse t tohef abli agsneesd itrna i(n2i)n. [sent-593, score-0.122]
51 xim", uthme likelihood estimation of these parameters in this model is intractable, efficient approximate learning of DBMs can be carried out by using mean-field inference together with Markov Chain Monte Carlo algorithms [14]. [sent-601, score-0.05]
52 Since the shapes often have similar local structural properties, the visible units can be divided into e)qual sized square patches to improve the learning p)ro? [sent-603, score-0.464]
53 are restricted so that they receive inputs only from a )squ? [sent-607, score-0.127]
54 In order to demonstrate )the advantages of three-layered DBM, we consider a simple case that the training shape is divided into four square patches for the arm posture experiment (Fig. [sent-609, score-0.29]
wordName wordTfidf (topN-words)
[('deep', 0.246), ('dbm', 0.224), ('boltzmann', 0.217), ('oaenrr', 0.205), ('shapes', 0.176), ('shape', 0.172), ('tve', 0.168), ('architecture', 0.144), ('units', 0.137), ('asphpapineg', 0.137), ('dleh', 0.137), ('energetic', 0.137), ('evolutional', 0.137), ('svtiastieb', 0.137), ('sdf', 0.106), ('ah', 0.105), ('ett', 0.101), ('bw', 0.101), ('ae', 0.098), ('variational', 0.094), ('rbm', 0.09), ('visible', 0.085), ('ao', 0.083), ('nt', 0.083), ('nv', 0.078), ('receive', 0.078), ('sa', 0.074), ('tah', 0.074), ('oo', 0.071), ('prior', 0.071), ('mc', 0.069), ('square', 0.066), ('ar', 0.064), ('pl', 0.062), ('dbms', 0.061), ('tohef', 0.061), ('shh', 0.061), ('itrna', 0.061), ('arr', 0.061), ('eag', 0.061), ('edl', 0.061), ('eyt', 0.061), ('hoaf', 0.061), ('nmt', 0.061), ('oafs', 0.061), ('ofb', 0.061), ('quest', 0.061), ('rsf', 0.061), ('squ', 0.061), ('sye', 0.061), ('tahrey', 0.061), ('tne', 0.061), ('twi', 0.061), ('china', 0.061), ('ob', 0.059), ('ht', 0.059), ('tio', 0.058), ('se', 0.057), ('rpa', 0.056), ('straint', 0.056), ('aenrde', 0.056), ('cedure', 0.056), ('cic', 0.056), ('cov', 0.056), ('egr', 0.056), ('erd', 0.056), ('ianb', 0.056), ('itso', 0.056), ('richly', 0.056), ('rtn', 0.056), ('stretching', 0.056), ('tee', 0.056), ('em', 0.056), ('woo', 0.053), ('tae', 0.053), ('nogf', 0.053), ('acoustic', 0.053), ('iti', 0.053), ('training', 0.052), ('bo', 0.051), ('sri', 0.05), ('embodied', 0.05), ('enr', 0.05), ('tit', 0.05), ('uthme', 0.05), ('afs', 0.05), ('aot', 0.05), ('hda', 0.05), ('hide', 0.05), ('oab', 0.05), ('ot', 0.05), ('fe', 0.05), ('guide', 0.05), ('inputs', 0.049), ('cope', 0.049), ('segmenting', 0.049), ('ti', 0.049), ('xim', 0.048), ('fei', 0.048), ('ica', 0.048)]
simIndex simValue paperId paperTitle
same-paper 1 1.0000001 105 cvpr-2013-Deep Learning Shape Priors for Object Segmentation
Author: Fei Chen, Huimin Yu, Roland Hu, Xunxun Zeng
Abstract: In this paper we introduce a new shape-driven approach for object segmentation. Given a training set of shapes, we first use deep Boltzmann machine to learn the hierarchical architecture of shape priors. This learned hierarchical architecture is then used to model shape variations of global and local structures in an energetic form. Finally, it is applied to data-driven variational methods to perform object extraction of corrupted data based on shape probabilistic representation. Experiments demonstrate that our model can be applied to dataset of arbitrary prior shapes, and can cope with image noise and clutter, as well as partial occlusions.
2 0.14098983 156 cvpr-2013-Exploring Compositional High Order Pattern Potentials for Structured Output Learning
Author: Yujia Li, Daniel Tarlow, Richard Zemel
Abstract: When modeling structured outputs such as image segmentations, prediction can be improved by accurately modeling structure present in the labels. A key challenge is developing tractable models that are able to capture complex high level structure like shape. In this work, we study the learning of a general class of pattern-like high order potential, which we call Compositional High Order Pattern Potentials (CHOPPs). We show that CHOPPs include the linear deviation pattern potentials of Rother et al. [26] and also Restricted Boltzmann Machines (RBMs); we also establish the near equivalence of these two models. Experimentally, we show that performance is affected significantly by the degree of variability present in the datasets, and we define a quantitative variability measure to aid in studying this. We then improve CHOPPs performance in high variability datasets with two primary contributions: (a) developing a loss-sensitive joint learning procedure, so that internal pattern parameters can be learned in conjunction with other model potentials to minimize expected loss;and (b) learning an image-dependent mapping that encourages or inhibits patterns depending on image features. We also explore varying how multiple patterns are composed, and learning convolutional patterns. Quantitative results on challenging highly variable datasets show that the joint learning and image-dependent high order potentials can improve performance.
3 0.13907923 50 cvpr-2013-Augmenting CRFs with Boltzmann Machine Shape Priors for Image Labeling
Author: Andrew Kae, Kihyuk Sohn, Honglak Lee, Erik Learned-Miller
Abstract: Conditional random fields (CRFs) provide powerful tools for building models to label image segments. They are particularly well-suited to modeling local interactions among adjacent regions (e.g., superpixels). However, CRFs are limited in dealing with complex, global (long-range) interactions between regions. Complementary to this, restricted Boltzmann machines (RBMs) can be used to model global shapes produced by segmentation models. In this work, we present a new model that uses the combined power of these two network types to build a state-of-the-art labeler. Although the CRF is a good baseline labeler, we show how an RBM can be added to the architecture to provide a global shape bias that complements the local modeling provided by the CRF. We demonstrate its labeling performance for the parts of complex face images from the Labeled Faces in the Wild data set. This hybrid model produces results that are both quantitatively and qualitatively better than the CRF alone. In addition to high-quality labeling results, we demonstrate that the hidden units in the RBM portion of our model can be interpreted as face attributes that have been learned without any attribute-level supervision.
Author: Roni Mittelman, Honglak Lee, Benjamin Kuipers, Silvio Savarese
Abstract: The use of semantic attributes in computer vision problems has been gaining increased popularity in recent years. Attributes provide an intermediate feature representation in between low-level features and the class categories, leading to improved learning on novel categories from few examples. However, a major caveat is that learning semantic attributes is a laborious task, requiring a significant amount of time and human intervention to provide labels. In order to address this issue, we propose a weakly supervised approach to learn mid-level features, where only class-level supervision is provided during training. We develop a novel extension of the restricted Boltzmann machine (RBM) by incorporating a Beta-Bernoulli process factor potential for hidden units. Unlike the standard RBM, our model uses the class labels to promote category-dependent sharing of learned features, which tends to improve the generalization performance. By using semantic attributes for which annotations are available, we show that we can find correspondences between the learned mid-level features and the labeled attributes. Therefore, the mid-level features have distinct semantic characterization which is similar to that given by the semantic attributes, even though their labeling was not provided during training. Our experimental results on object recognition tasks show significant performance gains, outperforming existing methods which rely on manually labeled semantic attributes.
5 0.12891772 253 cvpr-2013-Learning Multiple Non-linear Sub-spaces Using K-RBMs
Author: Siddhartha Chandra, Shailesh Kumar, C.V. Jawahar
Abstract: Understanding the nature of data is the key to building good representations. In domains such as natural images, the data comes from very complex distributions which are hard to capture. Feature learning intends to discover or best approximate these underlying distributions and use their knowledge to weed out irrelevant information, preserving most of the relevant information. Feature learning can thus be seen as a form of dimensionality reduction. In this paper, we describe a feature learning scheme for natural images. We hypothesize that image patches do not all come from the same distribution, they lie in multiple nonlinear subspaces. We propose a framework that uses K Restricted Boltzmann Machines (K-RBMS) to learn multiple non-linear subspaces in the raw image space. Projections of the image patches into these subspaces gives us features, which we use to build image representations. Our algorithm solves the coupled problem of finding the right non-linear subspaces in the input space and associating image patches with those subspaces in an iterative EM like algorithm to minimize the overall reconstruction error. Extensive empirical results over several popular image classification datasets show that representations based on our framework outperform the traditional feature representations such as the SIFT based Bag-of-Words (BoW) and convolutional deep belief networks.
6 0.12313238 306 cvpr-2013-Non-rigid Structure from Motion with Diffusion Maps Prior
7 0.10804088 33 cvpr-2013-Active Contours with Group Similarity
9 0.098641351 388 cvpr-2013-Semi-supervised Learning of Feature Hierarchies for Object Detection in a Video
10 0.097230338 288 cvpr-2013-Modeling Mutual Visibility Relationship in Pedestrian Detection
11 0.097069204 161 cvpr-2013-Facial Feature Tracking Under Varying Facial Expressions and Face Poses Based on Restricted Boltzmann Machines
12 0.078008264 111 cvpr-2013-Dense Reconstruction Using 3D Object Shape Priors
13 0.077270076 321 cvpr-2013-PDM-ENLOR: Learning Ensemble of Local PDM-Based Regressions
14 0.077090807 113 cvpr-2013-Dense Variational Reconstruction of Non-rigid Surfaces from Monocular Video
15 0.075400837 154 cvpr-2013-Explicit Occlusion Modeling for 3D Object Class Representations
16 0.075177297 104 cvpr-2013-Deep Convolutional Network Cascade for Facial Point Detection
17 0.072152145 32 cvpr-2013-Action Recognition by Hierarchical Sequence Summarization
18 0.067275524 304 cvpr-2013-Multipath Sparse Coding Using Hierarchical Matching Pursuit
19 0.067011058 110 cvpr-2013-Dense Object Reconstruction with Semantic Priors
20 0.062508211 381 cvpr-2013-Scene Parsing by Integrating Function, Geometry and Appearance Models
topicId topicWeight
[(0, 0.159), (1, 0.021), (2, -0.01), (3, 0.027), (4, 0.054), (5, -0.002), (6, 0.004), (7, 0.0), (8, -0.0), (9, -0.054), (10, 0.033), (11, -0.011), (12, -0.083), (13, -0.054), (14, 0.042), (15, 0.125), (16, -0.078), (17, 0.103), (18, 0.078), (19, 0.058), (20, 0.037), (21, -0.126), (22, 0.041), (23, -0.086), (24, -0.032), (25, -0.012), (26, 0.032), (27, -0.031), (28, 0.035), (29, 0.03), (30, 0.011), (31, -0.018), (32, -0.124), (33, 0.029), (34, 0.048), (35, 0.005), (36, 0.013), (37, -0.023), (38, 0.0), (39, -0.077), (40, 0.041), (41, 0.062), (42, 0.058), (43, -0.08), (44, 0.035), (45, 0.046), (46, 0.001), (47, 0.056), (48, -0.017), (49, 0.088)]
simIndex simValue paperId paperTitle
same-paper 1 0.91733712 105 cvpr-2013-Deep Learning Shape Priors for Object Segmentation
Author: Fei Chen, Huimin Yu, Roland Hu, Xunxun Zeng
Abstract: In this paper we introduce a new shape-driven approach for object segmentation. Given a training set of shapes, we first use deep Boltzmann machine to learn the hierarchical architecture of shape priors. This learned hierarchical architecture is then used to model shape variations of global and local structures in an energetic form. Finally, it is applied to data-driven variational methods to perform object extraction of corrupted data based on shape probabilistic representation. Experiments demonstrate that our model can be applied to dataset of arbitrary prior shapes, and can cope with image noise and clutter, as well as partial occlusions.
2 0.67418998 253 cvpr-2013-Learning Multiple Non-linear Sub-spaces Using K-RBMs
Author: Siddhartha Chandra, Shailesh Kumar, C.V. Jawahar
Abstract: Understanding the nature of data is the key to building good representations. In domains such as natural images, the data comes from very complex distributions which are hard to capture. Feature learning intends to discover or best approximate these underlying distributions and use their knowledge to weed out irrelevant information, preserving most of the relevant information. Feature learning can thus be seen as a form of dimensionality reduction. In this paper, we describe a feature learning scheme for natural images. We hypothesize that image patches do not all come from the same distribution, they lie in multiple nonlinear subspaces. We propose a framework that uses K Restricted Boltzmann Machines (K-RBMS) to learn multiple non-linear subspaces in the raw image space. Projections of the image patches into these subspaces gives us features, which we use to build image representations. Our algorithm solves the coupled problem of finding the right non-linear subspaces in the input space and associating image patches with those subspaces in an iterative EM like algorithm to minimize the overall reconstruction error. Extensive empirical results over several popular image classification datasets show that representations based on our framework outperform the traditional feature representations such as the SIFT based Bag-of-Words (BoW) and convolutional deep belief networks.
3 0.6613614 156 cvpr-2013-Exploring Compositional High Order Pattern Potentials for Structured Output Learning
Author: Yujia Li, Daniel Tarlow, Richard Zemel
Abstract: When modeling structured outputs such as image segmentations, prediction can be improved by accurately modeling structure present in the labels. A key challenge is developing tractable models that are able to capture complex high level structure like shape. In this work, we study the learning of a general class of pattern-like high order potential, which we call Compositional High Order Pattern Potentials (CHOPPs). We show that CHOPPs include the linear deviation pattern potentials of Rother et al. [26] and also Restricted Boltzmann Machines (RBMs); we also establish the near equivalence of these two models. Experimentally, we show that performance is affected significantly by the degree of variability present in the datasets, and we define a quantitative variability measure to aid in studying this. We then improve CHOPPs performance in high variability datasets with two primary contributions: (a) developing a loss-sensitive joint learning procedure, so that internal pattern parameters can be learned in conjunction with other model potentials to minimize expected loss;and (b) learning an image-dependent mapping that encourages or inhibits patterns depending on image features. We also explore varying how multiple patterns are composed, and learning convolutional patterns. Quantitative results on challenging highly variable datasets show that the joint learning and image-dependent high order potentials can improve performance.
Author: Roni Mittelman, Honglak Lee, Benjamin Kuipers, Silvio Savarese
Abstract: The use of semantic attributes in computer vision problems has been gaining increased popularity in recent years. Attributes provide an intermediate feature representation in between low-level features and the class categories, leading to improved learning on novel categories from few examples. However, a major caveat is that learning semantic attributes is a laborious task, requiring a significant amount of time and human intervention to provide labels. In order to address this issue, we propose a weakly supervised approach to learn mid-level features, where only class-level supervision is provided during training. We develop a novel extension of the restricted Boltzmann machine (RBM) by incorporating a Beta-Bernoulli process factor potential for hidden units. Unlike the standard RBM, our model uses the class labels to promote category-dependent sharing of learned features, which tends to improve the generalization performance. By using semantic attributes for which annotations are available, we show that we can find correspondences between the learned mid-level features and the labeled attributes. Therefore, the mid-level features have distinct semantic characterization which is similar to that given by the semantic attributes, even though their labeling was not provided during training. Our experimental results on object recognition tasks show significant performance gains, outperforming existing methods which rely on manually labeled semantic attributes.
5 0.65511084 50 cvpr-2013-Augmenting CRFs with Boltzmann Machine Shape Priors for Image Labeling
Author: Andrew Kae, Kihyuk Sohn, Honglak Lee, Erik Learned-Miller
Abstract: Conditional random fields (CRFs) provide powerful tools for building models to label image segments. They are particularly well-suited to modeling local interactions among adjacent regions (e.g., superpixels). However, CRFs are limited in dealing with complex, global (long-range) interactions between regions. Complementary to this, restricted Boltzmann machines (RBMs) can be used to model global shapes produced by segmentation models. In this work, we present a new model that uses the combined power of these two network types to build a state-of-the-art labeler. Although the CRF is a good baseline labeler, we show how an RBM can be added to the architecture to provide a global shape bias that complements the local modeling provided by the CRF. We demonstrate its labeling performance for the parts of complex face images from the Labeled Faces in the Wild data set. This hybrid model produces results that are both quantitatively and qualitatively better than the CRF alone. In addition to high-quality labeling results, we demonstrate that the hidden units in the RBM portion of our model can be interpreted as face attributes that have been learned without any attribute-level supervision.
6 0.61223221 32 cvpr-2013-Action Recognition by Hierarchical Sequence Summarization
7 0.56186861 33 cvpr-2013-Active Contours with Group Similarity
8 0.55606216 467 cvpr-2013-Wide-Baseline Hair Capture Using Strand-Based Refinement
9 0.55540031 321 cvpr-2013-PDM-ENLOR: Learning Ensemble of Local PDM-Based Regressions
10 0.54764128 371 cvpr-2013-SCaLE: Supervised and Cascaded Laplacian Eigenmaps for Visual Object Recognition Based on Nearest Neighbors
11 0.5405848 110 cvpr-2013-Dense Object Reconstruction with Semantic Priors
12 0.53668344 341 cvpr-2013-Procrustean Normal Distribution for Non-rigid Structure from Motion
13 0.51721823 306 cvpr-2013-Non-rigid Structure from Motion with Diffusion Maps Prior
14 0.51316607 218 cvpr-2013-Improving the Visual Comprehension of Point Sets
15 0.48328644 113 cvpr-2013-Dense Variational Reconstruction of Non-rigid Surfaces from Monocular Video
16 0.46569043 304 cvpr-2013-Multipath Sparse Coding Using Hierarchical Matching Pursuit
17 0.45867026 359 cvpr-2013-Robust Discriminative Response Map Fitting with Constrained Local Models
18 0.45512837 104 cvpr-2013-Deep Convolutional Network Cascade for Facial Point Detection
19 0.44043958 328 cvpr-2013-Pedestrian Detection with Unsupervised Multi-stage Feature Learning
20 0.43885994 353 cvpr-2013-Relative Hidden Markov Models for Evaluating Motion Skill
topicId topicWeight
[(10, 0.132), (16, 0.021), (26, 0.044), (33, 0.233), (49, 0.311), (67, 0.064), (69, 0.056), (87, 0.061)]
simIndex simValue paperId paperTitle
same-paper 1 0.77181131 105 cvpr-2013-Deep Learning Shape Priors for Object Segmentation
Author: Fei Chen, Huimin Yu, Roland Hu, Xunxun Zeng
Abstract: In this paper we introduce a new shape-driven approach for object segmentation. Given a training set of shapes, we first use deep Boltzmann machine to learn the hierarchical architecture of shape priors. This learned hierarchical architecture is then used to model shape variations of global and local structures in an energetic form. Finally, it is applied to data-driven variational methods to perform object extraction of corrupted data based on shape probabilistic representation. Experiments demonstrate that our model can be applied to dataset of arbitrary prior shapes, and can cope with image noise and clutter, as well as partial occlusions.
2 0.76937658 12 cvpr-2013-A Global Approach for the Detection of Vanishing Points and Mutually Orthogonal Vanishing Directions
Author: Michel Antunes, João P. Barreto
Abstract: This article presents a new global approach for detecting vanishing points and groups of mutually orthogonal vanishing directions using lines detected in images of man-made environments. These two multi-model fitting problems are respectively cast as Uncapacited Facility Location (UFL) and Hierarchical Facility Location (HFL) instances that are efficiently solved using a message passing inference algorithm. We also propose new functions for measuring the consistency between an edge and aputative vanishingpoint, and for computing the vanishing point defined by a subset of edges. Extensive experiments in both synthetic and real images show that our algorithms outperform the state-ofthe-art methods while keeping computation tractable. In addition, we show for the first time results in simultaneously detecting multiple Manhattan-world configurations that can either share one vanishing direction (Atlanta world) or be completely independent.
3 0.73785043 17 cvpr-2013-A Machine Learning Approach for Non-blind Image Deconvolution
Author: Christian J. Schuler, Harold Christopher Burger, Stefan Harmeling, Bernhard Schölkopf
Abstract: Image deconvolution is the ill-posed problem of recovering a sharp image, given a blurry one generated by a convolution. In this work, we deal with space-invariant non- blind deconvolution. Currently, the most successful meth- ods involve a regularized inversion of the blur in Fourier domain as a first step. This step amplifies and colors the noise, and corrupts the image information. In a second (and arguably more difficult) step, one then needs to remove the colored noise, typically using a cleverly engineered algorithm. However, the methods based on this two-step ap- proach do not properly address the fact that the image information has been corrupted. In this work, we also rely on a two-step procedure, but learn the second step on a large dataset of natural images, using a neural network. We will show that this approach outperforms the current state-ofthe-art on a large dataset of artificially blurred images. We demonstrate the practical applicability of our method in a real-world example with photographic out-of-focus blur.
4 0.69107181 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 0.67806929 248 cvpr-2013-Learning Collections of Part Models for Object Recognition
Author: Ian Endres, Kevin J. Shih, Johnston Jiaa, Derek Hoiem
Abstract: We propose a method to learn a diverse collection of discriminative parts from object bounding box annotations. Part detectors can be trained and applied individually, which simplifies learning and extension to new features or categories. We apply the parts to object category detection, pooling part detections within bottom-up proposed regions and using a boosted classifier with proposed sigmoid weak learners for scoring. On PASCAL VOC 2010, we evaluate the part detectors ’ ability to discriminate and localize annotated keypoints. Our detection system is competitive with the best-existing systems, outperforming other HOG-based detectors on the more deformable categories.
6 0.67521584 414 cvpr-2013-Structure Preserving Object Tracking
7 0.67208803 285 cvpr-2013-Minimum Uncertainty Gap for Robust Visual Tracking
8 0.67200285 325 cvpr-2013-Part Discovery from Partial Correspondence
9 0.67115587 408 cvpr-2013-Spatiotemporal Deformable Part Models for Action Detection
10 0.67024153 314 cvpr-2013-Online Object Tracking: A Benchmark
11 0.67004865 225 cvpr-2013-Integrating Grammar and Segmentation for Human Pose Estimation
12 0.66902912 446 cvpr-2013-Understanding Indoor Scenes Using 3D Geometric Phrases
13 0.66821808 104 cvpr-2013-Deep Convolutional Network Cascade for Facial Point Detection
14 0.66801929 445 cvpr-2013-Understanding Bayesian Rooms Using Composite 3D Object Models
15 0.66725761 365 cvpr-2013-Robust Real-Time Tracking of Multiple Objects by Volumetric Mass Densities
16 0.66712433 324 cvpr-2013-Part-Based Visual Tracking with Online Latent Structural Learning
17 0.66711396 70 cvpr-2013-Bottom-Up Segmentation for Top-Down Detection
18 0.66677821 14 cvpr-2013-A Joint Model for 2D and 3D Pose Estimation from a Single Image
19 0.66580468 74 cvpr-2013-CLAM: Coupled Localization and Mapping with Efficient Outlier Handling
20 0.66569078 267 cvpr-2013-Least Soft-Threshold Squares Tracking