Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Discriminative Pedestrian Features and Gated Channel Attention for Clothes-Changing Person Re-Identification

Yongkang Ding, Rui Mao, Hanyue Zhu, Anqi Wang, Liyan Zhang

TL;DR

This paper proposes an innovative method for disentangled feature extraction, effectively extracting discriminative features from pedestrian images that are invariant to clothing that leverages pedestrian parsing techniques to identify and retain features closely associated with individual identity while disregarding the variable nature of clothing attributes.

Abstract

In public safety and social life, the task of Clothes-Changing Person Re-Identification (CC-ReID) has become increasingly significant. However, this task faces considerable challenges due to appearance changes caused by clothing alterations. Addressing this issue, this paper proposes an innovative method for disentangled feature extraction, effectively extracting discriminative features from pedestrian images that are invariant to clothing. This method leverages pedestrian parsing techniques to identify and retain features closely associated with individual identity while disregarding the variable nature of clothing attributes. Furthermore, this study introduces a gated channel attention mechanism, which, by adjusting the network's focus, aids the model in more effectively learning and emphasizing features critical for pedestrian identity recognition. Extensive experiments conducted on two standard CC-ReID datasets validate the effectiveness of the proposed approach, with performance surpassing current leading solutions. The Top-1 accuracy under clothing change scenarios on the PRCC and VC-Clothes datasets reached 64.8% and 83.7%, respectively.

Discriminative Pedestrian Features and Gated Channel Attention for Clothes-Changing Person Re-Identification

TL;DR

This paper proposes an innovative method for disentangled feature extraction, effectively extracting discriminative features from pedestrian images that are invariant to clothing that leverages pedestrian parsing techniques to identify and retain features closely associated with individual identity while disregarding the variable nature of clothing attributes.

Abstract

In public safety and social life, the task of Clothes-Changing Person Re-Identification (CC-ReID) has become increasingly significant. However, this task faces considerable challenges due to appearance changes caused by clothing alterations. Addressing this issue, this paper proposes an innovative method for disentangled feature extraction, effectively extracting discriminative features from pedestrian images that are invariant to clothing. This method leverages pedestrian parsing techniques to identify and retain features closely associated with individual identity while disregarding the variable nature of clothing attributes. Furthermore, this study introduces a gated channel attention mechanism, which, by adjusting the network's focus, aids the model in more effectively learning and emphasizing features critical for pedestrian identity recognition. Extensive experiments conducted on two standard CC-ReID datasets validate the effectiveness of the proposed approach, with performance surpassing current leading solutions. The Top-1 accuracy under clothing change scenarios on the PRCC and VC-Clothes datasets reached 64.8% and 83.7%, respectively.

Paper Structure

This paper contains 14 sections, 9 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. The Proposed Method
  3. Controllable Disentanglement Feature Extraction Method (CDM)
  4. Gated Channel Attention Mechanism (GCA)
  5. Training Loss
  6. Experiments
  7. Datasets and Evaluation
  8. Implement Details
  9. Comparison With the State-of-The-Art Methods
  10. Ablation Experiments
  11. Analysis of Different Components
  12. Analysis of Hyperparameter
  13. Qualitative Analysis through Visualization
  14. Conclusion

Figures (4)

  • Figure 1: Flowchart of the ReID process. (a) Illustration of conventional ReID model. (b) Illustration of our CC-ReID model.
  • Figure 2: Pipeline of the Proposed Method. By extracting pedestrian discriminative features from semantic segmentation maps and incorporating a gated channel attention mechanism into the network, the performance of the model is further enhanced.
  • Figure 3: Impact of the hyper-parameter $\lambda_{1}$ and $\lambda_{2}$.
  • Figure 4: Visualization of heatmaps on PRCC (left) and VC-Clothes (right). Highlighted areas represent regions of heightened focus by the model.