首页--工业技术论文--自动化技术、计算机技术论文--计算技术、计算机技术论文--计算机的应用论文--信息处理(信息加工)论文--模式识别与装置论文

基于机器学习的排序优化及其在行人再识别中的应用研究

摘要第5-7页
ABSTRACT第7-9页
Chapter 1 Introduction第23-27页
    1.1 Introduction第23-24页
    1.2 Motivations第24-25页
    1.3 Contributions第25-26页
    1.4 Dissertation Organization第26-27页
Chapter 2 Literature Review第27-49页
    2.1 Introduction第27-29页
    2.2 Appearance-based Person Re-Id第29-39页
        2.2.1 Feature Representation第30-31页
        2.2.2 Feature Description第31-39页
            2.2.2.1 Scale-Invariant Feature Transform (SIFT) Based Color Descriptor第32页
            2.2.2.2 Affine Invariant Interest Point Detector (Schmid Filters)第32-33页
            2.2.2.3 Local Binary Patterns (LBP)第33-34页
            2.2.2.4 Speed Up Robust Features (SURF)第34页
            2.2.2.5 Histogram of Oriented Gradients (HOG)第34-36页
            2.2.2.6 The Pyramid of the Histogram of Oriented Gradients (PHOG)第36页
            2.2.2.7 Local Maximal Occurrence (LOMO)第36-38页
            2.2.2.8 Hexagonal-SIFT第38-39页
    2.3 Metric Learning Methods第39-42页
        2.3.1 Large Margin Nearest Neighbor (LMNN)第40页
        2.3.2 Information Theoretic Metric Learning (ITML)第40-41页
        2.3.3 Probabilistic Relative Distance Comparison (PRDC)第41页
        2.3.4 Large Scale Metric Learning from Equivalence Constraints (KISSME)第41页
        2.3.5 Cross-view Quadratic Discriminant Analysis (XQDA)第41-42页
        2.3.6 Symmetry-Driven Accumulation of Local Features (SDALF)第42页
    2.4 Post-Rank Optimization and Prioritization Methods第42-46页
        2.4.1 Soft Biometrics Attribute-based Re-ranking第43-44页
        2.4.2 Post-rank Optimization (POP)第44页
        2.4.3 Bi-directional Re-ranking第44-45页
        2.4.4 Saliency Re-ranking第45页
        2.4.5 Discriminative Content and Context Analysis (DCIA)第45-46页
    2.5 Evaluation Metrics第46页
    2.6 Discussions第46-48页
    本章小结第48-49页
Chapter 3 Pre-rank Prioritization第49-57页
    3.1 Introduction第49页
    3.2 System Overview第49-53页
        3.2.1 Person Tracking and Detection第50页
        3.2.2 Color-based Pre-ranking第50-51页
        3.2.3 Signature Generation and Training第51-53页
    3.3 Results and Discussions第53-55页
        3.3.1 Datasets第53页
        3.3.2 Feature Extraction and Evaluation Protocols第53-54页
        3.3.3 Comparison with Other Person Re-Id Methods第54-55页
    3.4 Summary第55-56页
    总结第56-57页
Chapter 4 Post-rank Optimization via Hypergraphs第57-79页
    4.1 Introduction: Post-rank Optimization and Prioritization第57-59页
    4.2 Hypergraph-based Post-rank Optimization第59-62页
        4.2.1 Motivations of Using Hypergraph第59-60页
        4.2.2 Basic Notations Used in Hypergraph第60-62页
    4.3 System Overview第62-68页
        4.3.1 The Rank List Refinement第62-64页
        4.3.2 Hypergraph Learning for Re-ranking第64-66页
        4.3.3 Weight Learning of Hyperedges第66-68页
    4.4 Experiments and Results第68-77页
        4.4.1 Datasets第68-69页
        4.4.2 Feature Extraction and Evaluation Protocols第69-70页
        4.4.3 Evaluation with State-of-the-art Post-ranking Approaches第70-74页
        4.4.4 Evaluation with State-of-the-art Ranking Approaches第74-77页
    4.5 Summary第77页
    总结第77-79页
Chapter 5 Multi-feature Fusion Based Rank Optimization第79-87页
    5.1 Introduction: Multi-feature Fusion第79-80页
    5.2 System Overview第80-82页
        5.2.1 Multi-feature Selection and Fusion第80-82页
        5.2.2 Low Dimensional Embedding第82页
        5.2.3 Image Tree-based Re-ranking第82页
    5.3 Experiments and Results第82-86页
        5.3.1 Datasets第83-84页
        5.3.2 Evaluation Setting第84页
        5.3.3 Comparison with Other Methods第84-85页
        5.3.4 Comparison with Single Features第85-86页
    5.4 Summary第86页
    总结第86-87页
Chapter 6 POP:System Design and Performance EvaluationConsiderations第87-97页
    6.1 Introduction: POP System Design第87-94页
        6.1.1 POP Methods: Key to Good Results/Prioritization第87-88页
        6.1.2 Feature/Descriptor Level Challenges第88-89页
        6.1.3 Baseline Method Selection第89-90页
        6.1.4 Benchmark Datasets第90-93页
        6.1.5 Experimental Setup and Evaluation Protocols第93-94页
    6.2 Summary第94-95页
    总结第95-97页
Chapter 7 Discussions第97-101页
    7.1 Concluding Remarks and Summary of Contributions第97页
    7.2 Future Work第97-98页
    7.1 总结与分析第98页
    7.2 展望第98-101页
Appendix A第101-105页
References第105-115页
ACKOWLEDGEMENT第115-117页
List of Publications第117页
    1. Journal Publications第117页
    2. Conference Publications第117页

论文共117页,点击 下载论文
上一篇:非线性系统的分层控制问题研究
下一篇:线性负虚系统概念的推广及其性质研究