Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring Stronger Transformer Representation Learning for Occluded Person Re-Identification

Zhangjian Ji, Donglin Cheng, Kai Feng

TL;DR

A novel self-supervision and supervision combining transformer-based person re-identification (ReID) framework, namely SSSC-TransReID, is proposed, which outperforms the state-of-the-art ReID methods by large margins on the mean average accuracy (mAP) and Rank-1 accuracy.

Abstract

Due to some complex factors (e.g., occlusion, pose variation and diverse camera perspectives), extracting stronger feature representation in person re-identification remains a challenging task. In this paper, we proposed a novel self-supervision and supervision combining transformer-based person re-identification framework, namely SSSC-TransReID. Different from the general transformer-based person re-identification models, we designed a self-supervised contrastive learning branch, which can enhance the feature representation for person re-identification without negative samples or additional pre-training. In order to train the contrastive learning branch, we also proposed a novel random rectangle mask strategy to simulate the occlusion in real scenes, so as to enhance the feature representation for occlusion. Finally, we utilized the joint-training loss function to integrate the advantages of supervised learning with ID tags and self-supervised contrastive learning without negative samples, which can reinforce the ability of our model to excavate stronger discriminative features, especially for occlusion. Extensive experimental results on several benchmark datasets show our proposed model obtains superior Re-ID performance consistently and outperforms the state-of-the-art ReID methods by large margins on the mean average accuracy (mAP) and Rank-1 accuracy.

Exploring Stronger Transformer Representation Learning for Occluded Person Re-Identification

TL;DR

A novel self-supervision and supervision combining transformer-based person re-identification (ReID) framework, namely SSSC-TransReID, is proposed, which outperforms the state-of-the-art ReID methods by large margins on the mean average accuracy (mAP) and Rank-1 accuracy.

Abstract

Due to some complex factors (e.g., occlusion, pose variation and diverse camera perspectives), extracting stronger feature representation in person re-identification remains a challenging task. In this paper, we proposed a novel self-supervision and supervision combining transformer-based person re-identification framework, namely SSSC-TransReID. Different from the general transformer-based person re-identification models, we designed a self-supervised contrastive learning branch, which can enhance the feature representation for person re-identification without negative samples or additional pre-training. In order to train the contrastive learning branch, we also proposed a novel random rectangle mask strategy to simulate the occlusion in real scenes, so as to enhance the feature representation for occlusion. Finally, we utilized the joint-training loss function to integrate the advantages of supervised learning with ID tags and self-supervised contrastive learning without negative samples, which can reinforce the ability of our model to excavate stronger discriminative features, especially for occlusion. Extensive experimental results on several benchmark datasets show our proposed model obtains superior Re-ID performance consistently and outperforms the state-of-the-art ReID methods by large margins on the mean average accuracy (mAP) and Rank-1 accuracy.

Paper Structure

This paper contains 16 sections, 4 equations, 5 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related works
  3. Occluded person re-identification
  4. Transformer-based person re-identification
  5. Self-supervised learning
  6. Methodology
  7. Data augmentation
  8. Dual branch architecture
  9. Joint-training loss combining self-supervised and supervised learning
  10. Experiments
  11. Datasets and evaluation metric
  12. Implementation details
  13. Comparison with state-of-the-art methods
  14. Visualize and analysis
  15. Ablation study
  16. ...and 1 more sections

Figures (5)

  • Figure 1: Visualization of the different data augmentation methods.
  • Figure 2: The overall framework of self-supervised and supervised combining transformer-based person re-identification.
  • Figure 3: Grad-CAM visualization of attention maps. (a) Input images, (b) TransReID without using sliding window, (c) TransReID with sliding window, (d) our proposed method
  • Figure 4: Analysis of the impact of different mask ratios for our data augmentation method proposed in subsection \ref{['sec:da']}.
  • Figure 5: Analysis of the impact of the hyper-parameter $\lambda$ of the total loss function.