Deep Learning for Person Re-Identification: A Survey and Outlook (2020-01-13T00:00:00.000000Z)

TL;DR

A powerful AGW baseline is designed, achieving state-of-the-art or at least comparable performance on twelve datasets for four different Re-ID tasks, and a new evaluation metric (mINP) is introduced, indicating the cost for finding all the correct matches, which provides an additional criteria to evaluate the Re- ID system for real applications.

Abstract

Person re-identification (Re-ID) aims at retrieving a person of interest across multiple non-overlapping cameras. With the advancement of deep neural networks and increasing demand of intelligent video surveillance, it has gained significantly increased interest in the computer vision community. By dissecting the involved components in developing a person Re-ID system, we categorize it into the closed-world and open-world settings. The widely studied closed-world setting is usually applied under various research-oriented assumptions, and has achieved inspiring success using deep learning techniques on a number of datasets. We first conduct a comprehensive overview with in-depth analysis for closed-world person Re-ID from three different perspectives, including deep feature representation learning, deep metric learning and ranking optimization. With the performance saturation under closed-world setting, the research focus for person Re-ID has recently shifted to the open-world setting, facing more challenging issues. This setting is closer to practical applications under specific scenarios. We summarize the open-world Re-ID in terms of five different aspects. By analyzing the advantages of existing methods, we design a powerful AGW baseline, achieving state-of-the-art or at least comparable performance on twelve datasets for four different Re-ID tasks. Meanwhile, we introduce a new evaluation metric (mINP) for person Re-ID, indicating the cost for finding all the correct matches, which provides an additional criteria to evaluate the Re-ID system for real applications. Finally, some important yet under-investigated open issues are discussed.

Authors

References242 items

A Survey of Open-World Person Re-Identification

Person Re-Identification by Contour Sketch Under Moderate Clothing Change

Locality Aware Appearance Metric for Multi-Target Multi-Camera Tracking

Momentum Contrast for Unsupervised Visual Representation Learning

Modality-aware Collaborative Learning for Visible Thermal Person Re-Identification

Illumination-Invariant Person Re-Identification

Deep Adversarial Graph Attention Convolution Network for Text-Based Person Search

Discriminative Feature Learning With Consistent Attention Regularization for Person Re-Identification

Self-Critical Attention Learning for Person Re-Identification

Unsupervised Graph Association for Person Re-Identification

Beyond Human Parts: Dual Part-Aligned Representations for Person Re-Identification

Unsupervised Person Re-Identification by Camera-Aware Similarity Consistency Learning

Instance-Guided Context Rendering for Cross-Domain Person Re-Identification

Pose-Guided Feature Alignment for Occluded Person Re-Identification

Robust Person Re-Identification by Modelling Feature Uncertainty

View Confusion Feature Learning for Person Re-Identification

MVP Matching: A Maximum-Value Perfect Matching for Mining Hard Samples, With Application to Person Re-Identification

RGB-Infrared Cross-Modality Person Re-Identification via Joint Pixel and Feature Alignment

Deep Reinforcement Active Learning for Human-in-the-Loop Person Re-Identification

Person Search by Text Attribute Query As Zero-Shot Learning

Co-Segmentation Inspired Attention Networks for Video-Based Person Re-Identification

Cross-Dataset Person Re-Identification via Unsupervised Pose Disentanglement and Adaptation

Re-ID Driven Localization Refinement for Person Search

Second-Order Non-Local Attention Networks for Person Re-Identification

Global-Local Temporal Representations for Video Person Re-Identification

advPattern: Physical-World Attacks on Deep Person Re-Identification via Adversarially Transformable Patterns

SBSGAN: Suppression of Inter-Domain Background Shift for Person Re-Identification

Mixed High-Order Attention Network for Person Re-Identification

Recover and Identify: A Generative Dual Model for Cross-Resolution Person Re-Identification

ABD-Net: Attentive but Diverse Person Re-Identification

Self-Training With Progressive Augmentation for Unsupervised Cross-Domain Person Re-Identification

HSME: Hypersphere Manifold Embedding for Visible Thermal Person Re-Identification

A Strong Baseline and Batch Normalization Neck for Deep Person Re-Identification

When Does Label Smoothing Help?

VRSTC: Occlusion-Free Video Person Re-Identification

Attribute-Driven Feature Disentangling and Temporal Aggregation for Video Person Re-Identification

Generalizable Person Re-Identification by Domain-Invariant Mapping Network

Patch-Based Discriminative Feature Learning for Unsupervised Person Re-Identification

Re-Ranking via Metric Fusion for Object Retrieval and Person Re-Identification

Towards Rich Feature Discovery With Class Activation Maps Augmentation for Person Re-Identification

Adaptive Transfer Network for Cross-Domain Person Re-Identification

Learning to Reduce Dual-Level Discrepancy for Infrared-Visible Person Re-Identification

Interaction-And-Aggregation Network for Person Re-Identification

AANet: Attribute Attention Network for Person Re-Identifications

Distilled Person Re-Identification: Towards a More Scalable System

Query-Guided End-To-End Person Search

Omni-Scale Feature Learning for Person Re-Identification

Deep Constrained Dominant Sets for Person Re-Identification

Joint Discriminative and Generative Learning for Person Re-Identification

Multi-Similarity Loss With General Pair Weighting for Deep Metric Learning

Foreground-Aware Pyramid Reconstruction for Alignment-Free Occluded Person Re-Identification

Weakly Supervised Person Re-Identification

Unsupervised Embedding Learning via Invariant and Spreading Instance Feature

A Novel Unsupervised Camera-Aware Domain Adaptation Framework for Person Re-Identification

Invariance Matters: Exemplar Memory for Domain Adaptive Person Re-Identification

Learning Context Graph for Person Search

Perceive Where to Focus: Learning Visibility-Aware Part-Level Features for Partial Person Re-Identification

Deep learning-based methods for person re-identification: A comprehensive review

Auto-ReID: Searching for a Part-Aware ConvNet for Person Re-Identification

Unsupervised Person Re-Identification by Soft Multilabel Learning

Unsupervised Person Re-Identification by Deep Asymmetric Metric Embedding

Dynamic Graph Co-Matching for Unsupervised Video-Based Person Re-Identification

Densely Semantically Aligned Person Re-Identification

Dissecting Person Re-Identification From the Viewpoint of Viewpoint

Spectral Feature Transformation for Person Re-Identification

Self-Similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-Identification

Re-Identification With Consistent Attentive Siamese Networks

Batch DropBlock Network for Person Re-Identification and Beyond

STA: Spatial-Temporal Attention for Large-Scale Video-based Person Re-Identification

Pyramidal Person Re-IDentification via Multi-Loss Dynamic Training

Heated-Up Softmax Embedding

Mancs: A Multi-task Attentional Network with Curriculum Sampling for Person Re-Identification

Robust Anchor Embedding for Unsupervised Video Person re-IDentification in the Wild

Deep Cross-Modal Projection Learning for Image-Text Matching

Unsupervised Person Re-identification by Deep Learning Tracklet Association

Generalizing a Person Retrieval Model Hetero- and Homogeneously

Improving Deep Visual Representation for Person Re-identification by Global and Local Image-language Association

Hard-Aware Point-to-Set Deep Metric for Person Re-identification

Adversarial Open-World Person Re-Identification

Survey on Deep Learning Techniques for Person Re-Identification Task

SphereReID: Deep Hypersphere Manifold Embedding for Person Re-Identification

Cascaded SR-GAN for Scale-Adaptive Low Resolution Person Re-identification

Visible Thermal Person Re-Identification via Dual-Constrained Top-Ranking

Person Re-identification with Cascaded Pairwise Convolutions

Adversarially Occluded Samples for Person Re-identification

Mask-Guided Contrastive Attention Model for Person Re-identification

RPIfield: A New Dataset for Temporally Evaluating Person Re-identification

Exploit the Unknown Gradually: One-Shot Video-Based Person Re-identification by Stepwise Learning

Group Consistent Similarity Learning via Deep CRF for Person Re-identification

Easy Identification from Better Constraints: Multi-shot Person Re-identification from Reference Constraints

Pose Transferrable Person Re-identification

End-to-End Deep Kronecker-Product Matching for Person Re-identification

Eliminating Background-bias for Robust Person Re-identification

Exploiting Transitivity for Learning Person Re-identification Models on a Budget

Video Person Re-identification with Competitive Snippet-Similarity Aggregation and Co-attentive Snippet Embedding

Resource Aware Person Re-identification Across Multiple Resolutions

Attention-Aware Compositional Network for Person Re-identification

Fast Open-World Person Re-Identification

Hierarchical Discriminative Learning for Visible Thermal Person Re-Identification

100

Domain Adaptation through Synthesis for Unsupervised Person Re-identification

101

Part-Aligned Bilinear Representations for Person Re-identification

102

Co-teaching: Robust training of deep neural networks with extremely noisy labels

103

Learning Discriminative Features with Multiple Granularities for Person Re-Identification

104

Human Semantic Parsing for Person Re-identification

105

Efficient and Deep Person Re-identification Using Multi-level Similarity

106

Features for Multi-target Multi-camera Tracking and Re-identification

107

Dual Attention Matching Network for Context-Aware Feature Sequence Based Person Re-identification

108

Diversity Regularized Spatiotemporal Attention for Video-Based Person Re-identification

109

Transferable Joint Attribute-Identity Deep Learning for Unsupervised Person Re-identification

110

Unsupervised Cross-Dataset Person Re-identification by Transfer Learning of Spatial-Temporal Patterns

111

Deep Cosine Metric Learning for Person Re-identification

112

Multi-level Factorisation Net for Person Re-identification

113

Video Person Re-Identification by Temporal Residual Learning

114

Harmonious Attention Network for Person Re-identification

115

Deep Spatial Feature Reconstruction for Partial Person Re-identification: Alignment-free Approach

116

Multi-shot Pedestrian Re-identification via Sequential Decision Making

117

Pose-Normalized Image Generation for Person Re-identification

118

A Pose-Sensitive Embedding for Person Re-identification with Expanded Cross Neighborhood Re-ranking

119

Camera Style Adaptation for Person Re-identification

120

Beyond Part Models: Person Retrieval with Refined Part Pooling

121

Person Transfer GAN to Bridge Domain Gap for Person Re-identification

122

Region-based Quality Estimation Network for Large-scale Person Re-identification

123

AlignedReID: Surpassing Human-Level Performance in Person Re-Identification

124

Non-local Neural Networks

125

Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identification

126

One-pass Person Re-identification by Sketch Online Discriminant Analysis

127

Fine-Tuning CNN Image Retrieval with No Human Annotation

128

Learning Deep Context-Aware Features over Body and Latent Parts for Person Re-identification

129

RGB-Infrared Cross-Modality Person Re-identification

130

Stepwise Metric Promotion for Unsupervised Video Person Re-identification

131

Efficient Online Local Metric Adaptation via Negative Samples for Person Re-identification

132

A Two Stream Siamese Convolutional Neural Network for Person Re-identification

133

Learning View-Invariant Features for Person Identification in Temporally Synchronized Videos Taken by Wearable Cameras

134

SHaPE: A Novel Graph Theoretic Algorithm for Making Consensus-Based Decisions in Person Re-identification Systems

135

Dynamic Label Graph Matching for Unsupervised Video Re-identification

136

Pose-Driven Deep Convolutional Model for Person Re-identification

137

Multi-scale Deep Learning Architectures for Person Re-identification

138

Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification

139

Random Erasing Data Augmentation

140

Identity-Aware Textual-Visual Matching with Latent Co-attention

141

Jointly Attentive Spatial-Temporal Pooling Networks for Video-Based Person Re-identification

142

Deeply-Learned Part-Aligned Representations for Person Re-identification

143

Consistent-Aware Deep Learning for Person Re-identification in a Camera Network

144

Neural Person Search Machines

145

Deep Representation Learning With Part Loss for Person Re-Identification

146

Fast Person Re-identification via Cross-Camera Semantic Binary Transformation

147

Point to Set Similarity Based Deep Feature Learning for Person Re-Identification

148

One-Shot Metric Learning for Person Re-identification

149

Spindle Net: Person Re-identification with Human Body Region Guided Feature Decomposition and Fusion

150

See the Forest for the Trees: Joint Spatial and Temporal Recurrent Neural Networks for Video-Based Person Re-identification

151

Multiple People Tracking by Lifted Multicut and Person Re-identification

152

Deep Reinforcement Learning Attention Selection For Person Re-Identification

153

Unsupervised Adaptive Re-identification in Open World Dynamic Camera Networks

154

Unsupervised Person Re-identification

155

Reinforced Temporal Attention and Split-Rate Transfer for Depth-Based Person Re-identification

156

Spatio-Temporal Person Retrieval via Natural Language Queries

157

Part-Based Deep Hashing for Large-Scale Person Re-Identification

158

Beyond Triplet Loss: A Deep Quadruplet Network for Person Re-identification

159

Robust Depth-Based Person Re-Identification

160

Scalable Person Re-identification on Supervised Smoothed Manifold

161

In Defense of the Triplet Loss for Person Re-Identification

162

Improving Person Re-identification by Attribute and Identity Learning

163

SVDNet for Pedestrian Retrieval

164

Person Recognition System Based on a Combination of Body Images from Visible Light and Thermal Cameras

165

An Overview and Empirical Comparison of Distance Metric Learning Methods

166

Person Search with Natural Language Description

167

Structured Deep Hashing with Convolutional Neural Networks for Fast Person Re-identification

168

Re-ranking Person Re-identification with k-Reciprocal Encoding

169

Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in Vitro

170

Looking beyond appearances: Synthetic training data for deep CNNs in re-identification

171

ArtTrack: Articulated Multi-Person Tracking in the Wild

172

Person Reidentification via Ranking Aggregation of Similarity Pulling and Dissimilarity Pushing

173

A Discriminatively Learned CNN Embedding for Person Reidentification

174

Person Re-identification: Past, Present and Future

175

MARS: A Video Benchmark for Large-Scale Person Re-Identification

176

Human-in-the-Loop Person Re-identification

177

Embedding Deep Metric for Person Re-identification: A Study Against Large Variations

178

Human Re-identification in Crowd Videos Using Personal, Social and Environmental Constraints

179

Person Re-identification via Recurrent Feature Aggregation

180

Towards unsupervised open-set person re-identification

181

A Siamese Long Short-Term Memory Architecture for Human Re-identification

182

Recurrent Convolutional Network for Video-Based Person Re-identification

183

Person Re-identification by Multi-Channel Parts-Based CNN with Improved Triplet Loss Function

184

Joint Learning of Single-Image and Cross-Image Representations for Person Re-identification

185

Improving Person Re-identification via Pose-Aware Multi-shot Matching

186

Recurrent Attention Models for Depth-Based Person Identification

187

Hierarchical Gaussian Descriptor for Person Re-identification

188

A Systematic Evaluation and Benchmark for Person Re-Identification: Features, Metrics, and Datasets

189

Deep Attributes Driven Multi-Camera Person Re-identification

190

Learning Deep Feature Representations with Domain Guided Dropout for Person Re-identification

191

Person Re-identification in the Wild

192

Joint Detection and Identification Feature Learning for Person Search

193

Towards Open-World Person Re-Identification by One-Shot Group-Based Verification

194

PersonNet: Person Re-identification with Deep Convolutional Neural Networks

195

Deep Residual Learning for Image Recognition

196

Scalable Person Re-identification: A Benchmark

197

Efficient PSD Constrained Asymmetric Metric Learning for Person Re-Identification

198

Partial Person Re-Identification

199

A Spatio-Temporal Appearance Representation for Video-Based Pedestrian Re-Identification

200

Multi-Scale Learning for Low-Resolution Person Re-Identification

201

Person Re-Identification with Discriminatively Trained Viewpoint Invariant Dictionaries

202

Ranking Optimization for Person Re-identification via Similarity and Dissimilarity

203

Specific Person Retrieval via Incomplete Text Description

204

An improved deep learning architecture for person re-identification

205

Query-adaptive late fusion for image search and person re-identification

206

Learning to rank in person re-identification with metric ensembles

207

Query Based Adaptive Re-ranking for Person Re-identification

208

Going deeper with convolutions

209

Person Re-identification by Video Ranking

210

Salient Color Names for Person Re-identification

211

Person Re-Identification Using Kernel-Based Metric Learning Methods

212

Saliency Weighted Features for Person Re-identification

213

Very Deep Convolutional Networks for Large-Scale Image Recognition

214

Deep Metric Learning for Person Re-identification

215

Improving person re-identification by viewpoint cues

216

DeepReID: Deep Filter Pairing Neural Network for Person Re-identification

217

Semi-supervised Coupled Dictionary Learning for Person Re-identification

218

Person re-identification by Local Maximal Occurrence representation and metric learning

219

A survey of approaches and trends in person re-identification

220

Domain Transfer Support Vector Ranking for Person Re-identification without Target Camera Label Information

221

POP: Person Re-identification Post-rank Optimisation

222

Person re-identification by manifold ranking

223

Unsupervised Salience Learning for Person Re-identification

224

ImageNet classification with deep convolutional neural networks

225

Relaxed Pairwise Learned Metric for Person Re-identification

226

Re-identification with RGB-D Sensors

227

Large scale metric learning from equivalence constraints

228

Person re-identification by probabilistic relative distance comparison

229

Person Re-identification by Descriptive and Discriminative Classification

230

Survey of Pedestrian Detection for Advanced Driver Assistance Systems

231

Person re-identification by symmetry-driven accumulation of local features

232

Pedestrian detection: A benchmark

233

Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features

234

Shape and Appearance Context Modeling

235

Person Reidentification Using Spatiotemporal Appearance

236

On the Error-Reject Trade-Off in Biometric Verification Systems

237

Learning Modality-Specific Representations for Visible-Infrared Person Re-Identification

238

Bi-Directional Center-Constrained Top-Ranking for Visible Thermal Person Re-Identification

239

Person re-identification using CNN features learned from combination of attributes

240

Speciﬁc Person Retrieval via Incomplete Text Description

241

Intelligent multi-camera video surveillance: A review

242