Computer Vision - ECCV 2008 : 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part I

The four-volume set comprising LNCS volumes 5302/5303/5304/5305 constitutes the refereed proceedings of the 10th European Conference on Computer Vision, ECCV 2008, held in Marseille, France, in October 2008. The 243 revised papers presented were carefully reviewed and selected from a total of 871 papers submitted. The four books cover the entire range of current issues in computer vision. The papers are organized in topical sections on recognition, stereo, people and face recognition, object tracking, matching, learning and features, MRFs, segmentation, computational photography and active reconstruction. Front Matter....Pages - Something Old, Something New, Something Borrowed, Something Blue....Pages 1-1 Learning to Localize Objects with Structured Output Regression....Pages 2-15 Beyond Nouns: Exploiting Prepositions and Comparative Adjectives for Learning Visual Classifiers....Pages 16-29 Learning Spatial Context: Using Stuff to Find Things....Pages 30-43 Segmentation and Recognition Using Structure from Motion Point Clouds....Pages 44-57 Keypoint Signatures for Fast Learning and Recognition....Pages 58-71 Active Matching....Pages 72-85 Towards Scalable Dataset Construction: An Active Learning Approach....Pages 86-98 GeoS: Geodesic Image Segmentation....Pages 99-112 Simultaneous Motion Detection and Background Reconstruction with a Mixed-State Conditional Markov Random Field....Pages 113-126 Semidefinite Programming Heuristics for Surface Reconstruction Ambiguities....Pages 127-140 Robust Optimal Pose Estimation....Pages 141-153 Learning to Recognize Activities from the Wrong View Point....Pages 154-166 Joint Parametric and Non-parametric Curve Evolution for Medical Image Segmentation....Pages 167-178 Localizing Objects with Smart Dictionaries....Pages 179-192 Weakly Supervised Object Localization with Stable Segmentations....Pages 193-207 A Perceptual Comparison of Distance Measures for Color Constancy Algorithms....Pages 208-221 Scale Invariant Action Recognition Using Compound Features Mined from Dense Spatio-temporal Corners....Pages 222-233 Semi-supervised On-Line Boosting for Robust Tracking....Pages 234-247 Reformulating and Optimizing the Mumford-Shah Functional on a Graph — A Faster, Lower Energy Solution....Pages 248-261 Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features....Pages 262-275 Perspective Nonrigid Shape and Motion Recovery....Pages 276-289 Shadows in Three-Source Photometric Stereo....Pages 290-303 Hamming Embedding and Weak Geometric Consistency for Large Scale Image Search....Pages 304-317 Estimating Geo-temporal Location of Stationary Cameras Using Shadow Trajectories....Pages 318-331 An Experimental Comparison of Discrete and Continuous Shape Optimization Methods....Pages 332-345 Image Feature Extraction Using Gradient Local Auto-Correlations....Pages 346-358 Analysis of Building Textures for Reconstructing Partially Occluded Facades....Pages 359-372 Nonrigid Image Registration Using Dynamic Higher-Order MRF Model....Pages 373-386 Tracking of Abrupt Motion Using Wang-Landau Monte Carlo Estimation....Pages 387-400 Surface Visibility Probabilities in 3D Cluttered Scenes....Pages 401-412 A Generative Shape Regularization Model for Robust Face Alignment....Pages 413-426 Modeling and Recognition of Landmark Image Collections Using Iconic Scene Graphs....Pages 427-440 VideoCut: Removing Irrelevant Frames by Discovering the Object of Interest....Pages 441-453 ASN: Image Keypoint Detection from Adaptive Shape Neighborhood....Pages 454-467 Online Sparse Matrix Gaussian Process Regression and Vision Applications....Pages 468-482 Multi-stage Contour Based Detection of Deformable Objects....Pages 483-496 Brain Hallucination....Pages 497-508 Range Flow for Varying Illumination....Pages 509-522 Some Objects Are More Equal Than Others: Measuring and Predicting Importance....Pages 523-536 Robust Multiple Structures Estimation with J-Linkage....Pages 537-547 Human Activity Recognition with Metric Learning....Pages 548-561 Shape Matching by Segmentation Averaging....Pages 562-575 Search Space Reduction for MRF Stereo....Pages 576-588 Estimating 3D Face Model and Facial Deformation from a Single Image Based on Expression Manifold Optimization....Pages 589-602 3D Face Recognition by Local Shape Difference Boosting....Pages 603-616 Efficiently Learning Random Fields for Stereo Vision with Sparse Message Passing....Pages 617-630 Recovering Light Directions and Camera Poses from a Single Sphere....Pages 631-642 Tracking with Dynamic Hidden-State Shape Models....Pages 643-656 Interactive Tracking of 2D Generic Objects with Spacetime Optimization....Pages 657-670 A Segmentation Based Variational Model for Accurate Optical Flow Estimation....Pages 671-684 Similarity Features for Facial Event Analysis....Pages 685-696 Building a Compact Relevant Sample Coverage for Relevance Feedback in Content-Based Image Retrieval....Pages 697-710 Discriminative Learning for Deformable Shape Segmentation: A Comparative Study....Pages 711-724 Discriminative Locality Alignment....Pages 725-738 Efficient Dense Scene Flow from Sparse or Dense Stereo Data....Pages 739-751 Integration of Multiview Stereo and Silhouettes Via Convex Functionals on Convex Domains....Pages 752-765 Using Multiple Hypotheses to Improve Depth-Maps for Multi-View Stereo....Pages 766-779 Sparse Structures in L-Infinity Norm Minimization for Structure and Motion Reconstruction....Pages 780-793 Back Matter....Pages - Welcome to the 2008EuropeanConference onComputer Vision. These proce- ings are the result of a great deal of hard work by many people. To produce them, a total of 871 papers were reviewed. Forty were selected for oral pres- tation and 203 were selected for poster presentation, yielding acceptance rates of 4.6% for oral, 23.3% for poster, and 27.9% in total. Weappliedthreeprinciples.First,sincewehadastronggroupofAreaChairs, the ?nal decisions to accept or reject a paper rested with the Area Chair, who wouldbeinformedbyreviewsandcouldactonlyinconsensuswithanotherArea Chair. Second, we felt that authors were entitled to a summary that explained how the Area Chair reached a decision for a paper. Third, we were very careful to avoid con?icts of interest. Each paper was assigned to an Area Chair by the Program Chairs, and each Area Chair received a pool of about 25 papers. The Area Chairs then identi?ed and rankedappropriatereviewersfor eachpaper in their pool, and a constrained optimization allocated three reviewers to each paper. We are very proud that every paper received at least three reviews. At this point, authors were able to respond to reviews. The Area Chairs then needed to reach a decision. We used a series of procedures to ensure careful review and to avoid con?icts of interest. ProgramChairs did not submit papers. The Area Chairs were divided into three groups so that no Area Chair in the group was in con?ict with any paper assigned to any Area Chair in the group. Annotation The four-volume set comprising LNCS volumes 5302/5303/5304/5305 constitutes the refereed proceedings of the 10th European Conference on Computer Vision, ECCV 2008, held in Marseille, France, in October 2008. The 243 revised papers presented were carefully reviewed and selected from a total of 871 papers submitted. The four books cover the entire range of current issues in computer vision. The papers are organized in topical sections on recognition, stereo, people and face recognition, object tracking, matching, learning and features, MRFs, segmentation, computational photography and active reconstruction Comprising of "LNCS" volumes 5302/5303/5304/5305, this set constitutes the refereed proceedings of the 10th European Conference on Computer Vision, ECCV 2008, held in Marseille, France, in October 2008. Organized in topical sections, it covers the entire range of issues in computer vision.