Chaohui Wang

Main menu:Home|Research|Teaching|Publications|Contact

1. Simultaneous segmentation, depth ordering and multi-object tracking

The goal of this project is to jointly and simultaneously solve segmentation, tracking and depth ordering in a single-shot optimization framework. To this end, we first introduce a joint 2.5D layered model where top-down object-level and bottom-up pixel-level representations are seamlessly combined through local constraints that involve only pairs of variables, as opposed to previous 2.5D layered models where the depth ordering was commonly modeled as a total and strict order between all the objects. Based on our layered modeling, we then propose a novel graphical-model formulation, where all the observed and hidden variables of interest such as image intensities, states of pixels (index of the associated object and relative depth) and of objects (motion parameters and relative depth) are jointly modeled within a single pairwise Markov random field (MRF). Finally, through minimizing the MRF energy, we simultaneously segment, track and sort by depth the objects.

C. Wang, M. de La Gorce, and N. Paragios
Segmentation, Ordering and Multi-object Tracking Using Graphical Models
In IEEE International Conference on Computer Vision (ICCV), 2009, Kyoto, Japan.

2. 3D surface analysis and registration

Similarity and correspondence are two fundamental archetype problems in shape analysis, encountered in numerous applications in computer vision and pattern recognition. We aim to advance the state of the art by searching for novel surface matching/registration formulations. We propose several different algorithms which achieve superior performance in challenging scenarios.

Y. Zeng, C. Wang, Y. Wang, X. Gu, D. Samaras, and N. Paragios
Dense Non-rigid Surface Registration Using High-order Graph Matching
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2010, San Francisco, USA.
C. Wang, M. M. Bronstein, A. M. Bronstein, and N. Paragios
Discrete Minimum Distortion Correspondence Problems for Non-rigid Shape Matching
International Conference on Scale Space and Variational Methods in Computer Vision (SSVM), 2011, The Dead Sea, Israel.
Y. Zeng, C. Wang, Y. Wang, X. Gu, D. Samaras, and N. Paragios
Intrinsic Dense 3D Surface Tracking
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2011, Colorado, USA.
Y. Zeng, C. Wang, Y. Wang, X. Gu, D. Samaras, and N. Paragios
A Generic Local Deformation Model for Shape Registration
Technical Report, INRIA, RR-7676, July, 2011.
C. Wang, Y. Zeng, D. Samaras, and N. Paragios
Modeling Shapes with Higher-Order Graphs: Methodology and Applications
Book Chapter in Shape Perception in Human and Computer Vision: An Interdisciplinary Perspective
(Editors: Sven Dickinson and Zygmunt Pizlo), Springer, 2013.
Y. Zeng, C. Wang, X. Gu, D. Samaras, and N. Paragios
Higher-order Graph Principles towards Non-rigid Surface Registration
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Accepted and to appear.

3. Shape modeling and knowledge-based image segmentation

We aim to advance the graph-based knowledge-based segmentation framework. On the one hand, regarding the prior model, we propose a novel higher-order Markov Random Field (MRF) model to encode pose-invariant priors and perform image segmentation. A dual-decomposition-based inference method is used to recover the optimal solution. On the other hand, regarding the regional data term, we propose an exact factorization of such a term in the MRF model using divergence theorem.

C. Wang, O. Teboul, F. Michel, S. Essafi, and N. Paragios
3D Knowledge-based Segmentation Using Pose-Invariant Higher-Order Graphs
In International Conference, Medical Image Computing and Computer Assisted Intervention (MICCAI), 2010, Beijing, China.
B. Xiang, C. Wang, J.-F. Deux, A. Rahmouni, and N. Paragios
Tagged Cardiac MR Image Segmentation Using Boundary & Regional-Support and Graph-based Deformable Priors
In IEEE International Symposium on Biomedical Imaging (ISBI), 2011, Chicago, USA.
B. Xiang, C. Wang, J.-F. Deux, A. Rahmouni, and N. Paragios
3D Cardiac Segmentation with Pose-Invariant Higher-Order MRFs
In IEEE International Symposium on Biomedical Imaging (ISBI), 2012, Barcelona, Spain.
C. Wang, Y. Zeng, D. Samaras, and N. Paragios
Modeling Shapes with Higher-Order Graphs: Methodology and Applications
Book Chapter in Shape Perception in Human and Computer Vision: An Interdisciplinary Perspective
(Editors: Sven Dickinson and Zygmunt Pizlo), Springer, 2013.

4. 3D Model Inference and Reconstruction From 2D Images

We are also interested in the inference and reconstuction of 3D model from 2D images. In particular, we aim to address the influence of camera pose in visual perception. To this end, we have proposed a novel one-shot optimization approach to simultaneously determine both the optimal 3D landmark model and the corresponding 2D projections from monocular 2D images without explicit estimation of the camera viewpoint, which is also able to deal with misdetections as well as partial occlusions.

C. Wang, Y. Zeng, L. Simon, I. Kakadiaris, D. Samaras and N. Paragios
Viewpoint Invariant 3D Landmark Model Inference from Monocular 2D Images Using Higher-Order Priors
In IEEE International Conference on Computer Vision (ICCV), 2011, Barcelona, Spain.
C. Wang, H. Boussaid, L. Simon, J.-Y. Lazennec and N. Paragios
Pose-invariant 3D Proximal Femur Estimation through Bi-Planar Image Segmentation with Hierarchical Higher-Order Graph-based Priors
In International Conference, Medical Image Computing and Computer Assisted Intervention (MICCAI), 2011, Toronto, Canada.

5. Bright channel cues and illumination estimation

We aim to infer the illumination environment and cast shadows from an observed image. On the one hand, we introduce a simple but efficient cue for the extraction of shadows from a single color image, the bright channel cue. On the other hand, we propose a higher-order Markov Random Field (MRF) illumination model to jointly recover the illumination environment and an estimate of the cast shadows in a scene from a single image, given coarse 3D geometry.

A. Panagopoulos, C. Wang, D. Samaras, and N. Paragios
Estimating Shadows with the Bright Channel Cue
In Color and Reflectance in Imaging and Computer Vision Workshop (CRICV)} (in conjuction with ECCV), 2010, Crete, Greece.
A. Panagopoulos, C. Wang, D. Samaras, and N. Paragios
Illumination Estimation and Cast Shadow Detection through a Higher-order Graphical Model
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2011, Colorado, USA.
A. Panagopoulos, C. Wang, D. Samaras, and N. Paragios
Simultaneous Cast Shadows, Illumination and Geometry Inference Using Hypergraphs
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Vol 35, Issue 2, Page 437-449, Feb. 2013.

6. Studies of modeling tools for computer vision

We are interested in studying and advancing important models for computer vision problems, such as MRF models and part-based models.

C. Wang, N. Komodakis, and N. Paragios
Markov Random Field Modeling, Inference & Learning in Computer Vision & Image Understanding: A Survey
Computer Vision and Image Understanding (CVIU), Vol 117, Issue 11, Page 1610-1627, Nov. 2013.
Y. Zeng, C. Wang, S. Soatto, and S. T. Yau
Nonlinearly Constrained MRFs: Exploring the Intrinsic Dimensions of Higher-Order Cliques
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013, Portland, USA.
Z. Ren, C. Wang, and A. Yuille
Scene-Domain Active Part Models for Object Representation
In IEEE International Conference on Computer Vision (ICCV), 2015, Santiago, Chile.

7. Indoor scene understanding

We aim to infer 3D objects and the layout of indoor scenes from a single RGB-D image captured with a Kinect camera. In particular, we have proposed a high-order graphical model for jointly reasoning about the layout, objects and superpixels in the image, which leverages detailed 3D geometry using inverse graphics and explicitly enforces occlusion and visibility constraints for respecting scene properties and projective geometry.

A. Geiger and C. Wang
Joint 3D Object and Layout Inference from a single RGB-D Image (*Best Paper Award*)
In German Conference on Pattern Recognition (GCPR), 2015, Aachen, Germany.

8. Single object tracking

We are interested in improving single object tracking techniques by adopting new strategies. Below are some methods that we have developed.

Z. Hong, C. Wang, X. Mei, D. Prokhorov, and D. Tao
Tracking using Multilevel Quantizations
In European Conference on Computer Vision (ECCV), 2014, Zürich, Switzerland..
Z. Hong, Z. Chen, C. Wang, X. Mei, D. Prokhorov, and D. Tao
MUlti-Store Tracker (MUSTer): a Cognitive Psychology Inspired Approach to Object Tracking
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015, Boston, USA.
X. Tian, L. Jiao, Z. Gan, C. Wang, and X. Zheng
Consistency-constrained Non-negative Coding for Tracking
IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), Accepted and to appear.