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

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

DOI: https://doi.org/10.1109/ICME57554.2024.10687558

2024-10-29

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.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper attempts to address the problem of recognition difficulties in the Person Re-Identification (ReID) task caused by pedestrians changing clothes. Specifically, traditional ReID methods usually rely on appearance features such as color, texture, and shape, but these features change significantly when pedestrians change clothes, reducing matching accuracy. Therefore, this paper proposes an innovative method aimed at extracting clothing-independent pedestrian features to improve person re-identification performance in scenarios where clothing changes.

### Main Contributions of the Paper:
1. **Controllable Decoupling Feature Extraction Method**: By generating semantic segmentation maps through pedestrian parsing technology and using a controllable decoupling method to convert these segmentation maps into clothing-independent grayscale images, clothing-independent pedestrian features are extracted.
2. **Gated Channel Attention Mechanism**: A gated channel attention mechanism is introduced to enhance the model's ability to learn key features by adjusting the network's focus on different feature channels.
3. **Two-Stage Optimization Strategy**: A two-stage optimization strategy is adopted during training, first using an identity loss function for basic feature learning, and then combining a triplet loss function to further improve the model's performance in complex scenarios.
4. **Experimental Validation**: Extensive experiments were conducted on two standard clothing change person re-identification datasets (PRCC and VC-Clothes). The results show that the proposed method achieved Top-1 accuracy rates of 64.8% and 83.7% respectively in clothing change scenarios, surpassing current leading methods.

### Background and Significance of the Paper:
With the rapid development of computer vision technology, person re-identification has become an important research field. Especially in public safety and social life, clothing change person re-identification (CC-ReID) is becoming increasingly important. However, this task faces significant challenges because the appearance of pedestrians changes significantly due to clothing changes. The innovative method proposed in this paper not only improves the recognition accuracy of the model but also enhances the robustness of the model in complex scenarios, with broad application prospects.