Human centric visual analysis with deep learning

Foreword......Page 5
Preface......Page 7
Contents......Page 9
Part I Motivation and Overview......Page 13
  1.1.1 Perceptron......Page 14
  1.1.2 Multilayer Perceptron......Page 15
  1.1.3 Formulation of Neural Network......Page 17
  1.2.1 Batch Normalization......Page 18
  1.2.2 Batch Kalman Normalization......Page 20
 References......Page 24
 2.1 Face Detection......Page 25
  2.2.1 Conventional Approaches......Page 26
  2.2.2 Deep-Learning-Based Models......Page 27
  2.3.1 Benchmarks for Pedestrian Detection......Page 28
  2.3.2 Pedestrian Detection Methods......Page 29
 References......Page 31
Part II Localizing Persons in Images......Page 36
 3.1 Facial Landmark Machines......Page 38
 3.2 The Cascaded BB-FCN Architecture......Page 40
  3.2.1 Backbone Network......Page 41
  3.2.2 Branch Network......Page 42
  3.2.3 Ground Truth Heat Map Generation......Page 43
  3.3.2 Evaluation Metric......Page 44
  3.3.4 Comparison with the State of the Art......Page 45
 3.4 Attention-Aware Face Hallucination......Page 46
  3.4.1 The Framework of Attention-Aware Face Hallucination......Page 48
  3.4.2 Recurrent Policy Network......Page 49
  3.4.4 Deep Reinforcement Learning......Page 51
  3.4.5 Experiments......Page 52
 References......Page 53
 4.1 Introduction......Page 55
  4.2.1 Region Proposal Network for Pedestrian Detection......Page 57
  4.2.3 Boosted Forest......Page 58
 4.3 Experiments and Analysis......Page 59
 References......Page 61
Part III Parsing Person in Detail......Page 63
 5.1 Introduction......Page 66
 5.2 Look into Person Benchmark......Page 68
 5.3 Self-supervised Structure-Sensitive Learning......Page 69
  5.3.1 Self-supervised Structure-Sensitive Loss......Page 71
  5.3.2 Experimental Result......Page 73
 References......Page 74
 6.1 Introduction......Page 76
 6.2 Related Work......Page 79
 6.3 Crowd Instance-Level Human Parsing Dataset......Page 80
  6.3.2 Dataset Statistics......Page 81
 6.4 Part Grouping Network......Page 82
  6.4.1 PGN Architecture......Page 83
  6.4.2 Instance Partition Process......Page 85
  6.5.1 Experimental Settings......Page 86
  6.5.2 PASCAL-Person-Part Dataset......Page 87
  6.5.4 Qualitative Results......Page 88
 References......Page 89
 7.1 Introduction......Page 91
 7.2 Video Instance-Level Parsing Dataset......Page 92
 7.3 Adaptive Temporal Encoding Network......Page 93
  7.3.2 Parsing R-CNN......Page 96
  7.3.3 Training and Inference......Page 97
 References......Page 99
Part IV Identifying and Verifying Persons......Page 100
 8.1 Introduction......Page 104
 8.2 Generalized Similarity Measures......Page 106
  8.2.1 Model Formulation......Page 109
  8.2.2 Connection with Existing Models......Page 110
  8.3.1 Deep Architecture......Page 111
  8.3.2 Model Training......Page 113
 8.4 Experiments......Page 116
 References......Page 117
 9.1 Introduction......Page 120
 9.2 Related Work......Page 122
 9.3 Framework Overview......Page 125
 9.4 Formulation and Optimization......Page 126
 References......Page 134
Part V Higher Level Tasks......Page 136
 10.1 Introduction......Page 139
 10.2 Deep Structured Model......Page 140
  10.2.2 Latent Temporal Structure......Page 141
  10.2.3 Deep Model with Relaxed Radius-Margin Bound......Page 143
  10.3.1 Latent Temporal Structure......Page 146
  10.3.2 Architecture of Deep Neural Networks......Page 147
 10.4 Learning Algorithm......Page 149
  10.4.1 Joint Component Learning......Page 150
  10.4.3 Inference......Page 153
  10.5.1 Datasets and Setting......Page 154
  10.5.2 Empirical Analysis......Page 155
 References......Page 159