Fine-Grained Image Analysis: Modern Approaches

Description

This book provides a comprehensive overview of the fine-grained image analysis research and modern approaches based on deep learning, spanning the full range of topics needed for designing operational fine-grained image systems. The author begins by providing detailed background information on FGIA, focusing on recognition and retrieval. The author also provides the fundamentals of convolutional neural networks to further make it easier for readers to understand the technical content in the book. The book introduces the main technical paradigms, technological developments, and representative approaches of fine-grained image recognition and fine-grained image retrieval. The author covers multiple popular research topics and includes cross-domain knowledge. The book also highlights advanced applications and topics for future research.   Foreword Preface Contents 1 Introduction 2 Background 2.1 Problem and Challenges 2.2 Recognition Versus Retrieval 2.3 Domain-Specific Applications Related to Fine-Grained Image Analysis 3 Benchmark Datasets 3.1 Introduction 3.2 Fine-Grained Recognition Datasets 3.2.1 CUB200-2011 3.2.2 Stanford Dogs 3.2.3 Stanford Cars 3.2.4 Oxford Flowers 3.2.5 iNaturalist 3.2.6 RPC 3.3 Fine-Grained Retrieval Datasets 3.3.1 DeepFashion 3.3.2 SBIR 3.3.3 QMUL Datasets 3.3.4 FG-Xmedia 4 Fine-Grained Image Recognition 4.1 Introduction 4.2 Recognition by Localization-Classification Subnetworks 4.2.1 Employing Detection or Segmentation Techniques 4.2.2 Utilizing Deep Filters 4.2.3 Leveraging Attention Mechanisms 4.2.4 Other Methods 4.3 Recognition by End-to-End Feature Encoding 4.3.1 Performing High-Order Feature Interactions 4.3.2 Designing Specific Loss Functions 4.3.3 Other Methods 4.4 Recognition with External Information 4.4.1 Noisy Web Data 4.4.2 Multi-Modal Data 4.4.3 Humans-in-the-Loop 4.5 Summary 5 Fine-Grained Image Retrieval 5.1 Introduction 5.2 Content-Based Fine-Grained Image Retrieval 5.2.1 Selective Convolutional Descriptor Aggregation 5.2.2 Centralized Ranking Loss 5.2.3 Category-Specific Nuance Exploration Network 5.3 Sketch-Based Fine-Grained Image Retrieval 5.3.1 ``Sketch Me That Shoe'' 5.3.2 Generalizing Fine-Grained Sketch-Based Image Retrieval 5.3.3 Jigsaw Puzzle for Fine-Grained Sketch-Based Image Retrieval 5.4 Summary 6 Resouces and Future Work 6.1 Deep Learning-Based Toolboxes 6.1.1 Hawkeye for Fine-Grained Recognition 6.1.2 PyRetri for Fine-Grained Retrieval 6.2 Conclusion Remarks and Future Directions A Vector, Matrix and Their Basic Operations A.1 Vector and Its Operations A.1.1 Vector A.1.2 Vector Norm A.1.3 Vector Operation A.2 Matrix and Its Operations A.2.1 Matrix A.2.2 Matrix Norm A.2.3 Matrix Operation Stochastic Gradient Descent Chain Rule Convolutional Neural Networks D.1 Development History D.2 Basic Structure D.3 Feed-Forward Operations D.4 Feed-Back Operations D.5 Basic Operations in CNNs D.5.1 Convolution Layers D.5.2 Pooling Layers D.5.3 Activation Functions D.5.4 Fully Connected Layers D.5.5 Objective Functions References Index