Person Re-Identification

Part I. Introduction --
1. About Behaviour --
1.1. Understanding Behaviour --
1.1.1. Representation and Modelling --
1.1.2. Detectionand Classification --
1.1.3. Predictionand Association --
1.2. Opportunities --
1.2.1. Visual Surveillance --
1.2.2. Video Indexing and Search --
1.2.3. Robotics and Healthcare --
1.2.4. Interaction,Animationand Computer Games --
1.3. Challenges --
1.3.1. Complexity --
1.3.2. Uncertainty --
1.4. The Approach --
2. Behaviour in Context --
2.1. Facial Expression --
2.2. Body Gesture --
2.3. Human Action --
2.4. Human Intent --
2.5. Group Activity --
2.6. Crowd Behaviour --
2.7. Distributed Behaviour --
2.8. Holistic Awareness: Connecting the Dots --
3. Towards Modelling Behaviour --
3.1. Behaviour Representation --
3.1.1. Object-Based Representation --
3.1.2. Part-Based Representation --
3.1.3. Pixel-Based Representation --
3.1.4. Event-Based Representation --
3.2. Probabilistic Graphical Models --
3.2.1. Static Bayesian Networks --
3.2.2. Dynamic Bayesian Networks --
3.2.3. Probabilistic Topic Models --
3.3. Learning Strategies --
3.3.1. Supervised Learning --
3.3.2. Unsupervised Learning --
3.3.3. Semi-supervised Learning --
3.3.4. Weakly Supervised Learning --
3.3.5. Active Learning --
Part II. Single-Object Behaviour --
4. Understanding Facial Expression --
4.1. Classification of Images --
4.1.1. Local Binary Patterns --
4.1.2. Designing Classifiers --
4.1.3. Feature Selection by Boosting --
4.2. Manifold and Temporal Modelling --
4.2.1. Locality Preserving Projections --
4.2.2. Bayesian Temporal Models --
4.3. Discussion --
5. Modelling Gesture --
5.1. Tracking Gesture --
5.1.1. Motion Moment Trajectory --
5.1.2. 2D Colour-Based Tracking --
5.1.3. Bayesian Association --
5.1.4. 3D Model-Based Tracking --
5.2. Segmentationand Atomic Action --
5.2.1. Temporal Segmentation --
5.2.2. Atomic Actions --
5.3. Markov Processes --
5.4. Affective State Analysis --
5.4.1. Space-Time Interest Points --
5.4.2. Expressionand Gesture Correlation --
5.5. Discussion --
6. Action Recognition --
6.1. Human Silhouette --
6.2. Hidden Conditional Random Fields --
6.2.1. HCRF Potential Function --
6.2.2. Observable HCRF --
6.3. Space-Time Clouds --
6.3.1. Clouds of Space-Time Interest Points --
6.3.2. Joint Local and Global Feature Representation --
6.4. Localisationand Detection --
6.4.1. Tracking Salient Points --
6.4.2. Automated Annotation --
6.5. Discussion --
Part III. Group Behaviour --
7. Supervised Learning of Group Activity --
7.1. Contextual Events --
7.1.1. Seeding Event: Measuring Pixel-Change-History --
7.1.2. Classificationof Contextual Events --
7.2. Activity Segmentation --
7.2.1. Semantic Content Extraction --
7.2.2. Semantic Video Segmentation --
7.3. Dynamic Bayesian Networks --
7.3.1. Correlationsof Temporal Processes --
7.3.2. Behavioural Interpretation of Activities --
7.4. Discussion --
8. Unsupervised Behaviour Profiling --
8.1. Off-line Behaviour Profile Discovery --
8.1.1. Behaviour Pattern --
8.1.2. Behaviour Profilingby Data Mining --
8.1.3. Behaviour Affinity Matrix --
8.1.4. Eigendecomposition --
8.1.5. Model Order Selection --
8.1.6. Quantifying Eigenvector Relevance --
8.2. On-line Anomaly Detection --
8.2.1. A Composite Behaviour Model --
8.2.2. Run-Time Anomaly Measure --
8.2.3. On-line Likelihood Ratio Test --
8.3. On-line Incremental Behaviour Modelling --
8.3.1. Model Bootstrapping --
8.3.2. Incremental Parameter Update --
8.3.3. Model Structure Adaptation --
8.4. Discussion --
9. Hierarchical Behaviour Discovery --
9.1. Local Motion Events --
9.2. Markov Clustering Topic Model --
9.2.1. Off-line Model Learning by Gibbs Sampling --
9.2.2. On-line Video Saliency Inference --
9.3. On-line Video Screening --
9.4. Model Complexity Control --
9.5. Semi-supervised Learning of Behavioural Saliency --
9.6. Discussion --
10. Learning Behavioural Context --
10.1. Spatial Context --
10.1.1. Behaviour-Footprint --
10.1.2. Semantic Scene Decomposition --
10.2. Correlational and Temporal Context --
10.2.1. Learning Regional Context --
10.2.2. Learning Global Context --
10.3. Context-Aware Anomaly Detection --
10.4. Discussion --
11. Modelling Rare and Subtle Behaviours --
11.1. Weakly Supervised Joint Topic Model --
11.1.1. Model Structure --
11.1.2. Model Parameters --
11.2. On-line Behaviour Classification --
11.3. Localisationof Rare Behaviour --
11.4. Discussion --
12. Man in the Loop --
12.1. Active Behaviour Learning Strategy --
12.2. Local Block-Based Behaviour --
12.3. Bayesian Classification --
12.4. Query Criteria --
12.4.1. Likelihood Criterion --
12.4.2. Uncertainty Criterion --
12.5. Adaptive Query Selection --
12.6. Discussion --
Part IV. Distributed Behaviour --
13. Multi-camera Behaviour Correlation --
13.1. Multi-view Activity Representation --
13.1.1. Local Bivariate Time-Series Events --
13.1.2. Activity-Based Scene Decomposition --
13.2. Learning Pair-Wise Correlation --
13.2.1. Cross Canonical Correlation Analysis --
13.2.2. Time-Delayed Mutual Information Analysis --
13.3. Multi-camera Topology Inference --
13.4. Discussion --
14. Person Re-identification --
14.1. Re-identification by Ranking --
14.1.1. Support Vector Ranking --
14.1.2. Scalability and Complexity --
14.1.3. Ensemble Rank SVM --
14.2. Context-Aware Search --
14.3. Discussion --
15. Connecting the Dots --
15.1. Global Behaviour Segmentation --
15.2. Bayesian Behaviour Graphs --
15.2.1. A Time-Delayed Probabilistic Graphical Model --
15.2.2. Bayesian Graph Structure Learning --
15.2.3. Bayesian Graph Parameter Learning --
15.2.4. Cumulative Anomaly Score --
15.2.5. Incremental Model Structure Learning --
15.3. Global Awareness --
15.3.1. Time-Ordered Latent Dirichlet Allocation --
15.3.2. On-line Predictionand Anomaly Detection --
15.4. Discussion.