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

基于弱监督学习的图像协同分割与定位

ABSTRACT第5-7页
摘要第8-14页
SYMBOL LIST第14-15页
ABBREVIATION LIST第15-21页
Chapter 1 Thesis Overview and Background第21-63页
    1.1 Motivation and Statement第21-22页
    1.2 Eye Localization第22-23页
    1.3 Image Cosegmentation第23-43页
    1.4 video cosegmentation第43-58页
    1.5 Main Results and Organization第58-63页
Chapter 2 Eye localization with Rotation Invariance第63-87页
    2.1 Introduction and Motivation第63-67页
    2.2 Rotation Invariant Eye Localization第67-74页
        2.2.1 Codebook of Invariant Local Features第67-69页
        2.2.2 2-class SRC with Multi-valued Output for Localization第69-71页
        2.2.3 Pyramid-like Detection and Localization第71-74页
    2.3 Experiments and Discussions第74-82页
        2.3.1 Localization within Local Search Regions第77-79页
        2.3.2 Localization with Face Detector Only第79-81页
        2.3.3 Localization without Face Detector第81-82页
    2.4 Conclusion第82-87页
Chapter 3 Image Cosegmentation Based on Mutual Learning Between Saliency andSimilarity第87-109页
    3.1 Introduction and Motivation第87-95页
        3.1.1 The Saliency of Common Objects第87-90页
        3.1.2 The Similarity of Common Objects第90-92页
        3.1.3 Preliminaries: Low-rank based Saliency Detection第92-95页
    3.2 Discriminative Learning based Mutual Learning Framework第95-99页
    3.3 Tree-Graph Cut based Mutual Learning Framework第99-102页
    3.4 Experiments and Discussions第102-108页
        3.4.1 Comparison with Other Co-segmentation Methods第103-106页
        3.4.2 Iteration Analysis第106-108页
    3.5 Conclusion第108-109页
Chapter 4 A Unified Mutual Learning Framework for Efficient Image Cosegmentation第109-141页
    4.1 Introduction and Motivation第109-110页
    4.2 A Unified Framework Based on Structured Sparsity and Tree-graph Matching第110-117页
        4.2.1 Preliminaries: The Rethinking of Tree-graph Matching第110-111页
        4.2.2 Proposed Model第111-114页
        4.2.3 Optimization第114-117页
    4.3 Two Strategies for Efficiency第117-120页
        4.3.1 Active Node Strategy第117-119页
        4.3.2 Self-adaptive Tree Reconstruction Strategy第119-120页
    4.4 Experiments and Discussions第120-133页
        4.4.1 The Performances of the Heat Maps第122-124页
        4.4.2 The Performances of cosegmentation第124-131页
        4.4.3 Parameter and Iteration Analysis第131-132页
        4.4.4 Discussion第132-133页
    4.5 Conclusion第133-137页
    4.6 Appendix第137-141页
        4.6.1 Updating L~i, S~i, α~i第137-140页
        4.6.2 Updating Weights第140-141页
Chapter 5 Video Cosegmentation with Deep Descriptors第141-149页
    5.1 Introduction and Motivation第141-142页
    5.2 Video cosegmentation with the Reuse of RNN Model第142-146页
        5.2.1 Preliminaries: Conditional Random Field Model第142-143页
        5.2.2 Pre-trained CRF-RNN for Feature Extraction第143-145页
        5.2.3 Clustering-based Video Cosegmentation with Deep Descriptors第145-146页
        5.2.4 Surgery on Pre-trained CRF-RNN for Refinements第146页
    5.3 Experiments and Discussion第146-148页
    5.4 Conclusion第148-149页
Chapter 6 Concluding Remarks and Future Work第149-155页
    6.1 Conclusions第149-150页
    6.2 Future Directions第150-155页
BIBLIOGRAPHY第155-175页
ACKNOWLEDGEMENTS第175-177页
RESUME第177-178页

论文共178页,点击 下载论文
上一篇:犬血清白蛋白融合干扰素γ的制备及功能研究
下一篇:青海省高寒草地退化综合评价研究