Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SiCL: Silhouette-Driven Contrastive Learning for Unsupervised Person Re-Identification with Clothes Change

Mingkun Li, Peng Xu, Chun-Guang Li, Jun Guo

TL;DR

This work tackles unsupervised long-term person re-identification under clothes change by introducing SiCL, a silhouette-driven contrastive learning framework that jointly leverages RGB imagery and person silhouettes. SiCL builds a two-branch network and a hierarchical fusion clustering strategy to establish garment-invariant representations, supplemented by prototypical, cross-view, and cluster-level neighbor contrastive losses and a curriculum learning scheme for neighbor selection. The approach achieves strong results on six clothes-change datasets, surpassing unsupervised short-term baselines and approaching the performance of some supervised long-term methods, while also providing a comprehensive benchmark and analysis. Overall, SiCL demonstrates the value of incorporating silhouette information and multi-level neighborhood structures to address cross-clothes invariance in unsupervised long-term re-ID, with potential impact on open-world re-ID systems and benchmarking.

Abstract

In this paper, we address a highly challenging yet critical task: unsupervised long-term person re-identification with clothes change. Existing unsupervised person re-id methods are mainly designed for short-term scenarios and usually rely on RGB cues so that fail to perceive feature patterns that are independent of the clothes. To crack this bottleneck, we propose a silhouette-driven contrastive learning (SiCL) method, which is designed to learn cross-clothes invariance by integrating both the RGB cues and the silhouette information within a contrastive learning framework. To our knowledge, this is the first tailor-made framework for unsupervised long-term clothes change \reid{}, with superior performance on six benchmark datasets. We conduct extensive experiments to evaluate our proposed SiCL compared to the state-of-the-art unsupervised person reid methods across all the representative datasets. Experimental results demonstrate that our proposed SiCL significantly outperforms other unsupervised re-id methods.

SiCL: Silhouette-Driven Contrastive Learning for Unsupervised Person Re-Identification with Clothes Change

TL;DR

This work tackles unsupervised long-term person re-identification under clothes change by introducing SiCL, a silhouette-driven contrastive learning framework that jointly leverages RGB imagery and person silhouettes. SiCL builds a two-branch network and a hierarchical fusion clustering strategy to establish garment-invariant representations, supplemented by prototypical, cross-view, and cluster-level neighbor contrastive losses and a curriculum learning scheme for neighbor selection. The approach achieves strong results on six clothes-change datasets, surpassing unsupervised short-term baselines and approaching the performance of some supervised long-term methods, while also providing a comprehensive benchmark and analysis. Overall, SiCL demonstrates the value of incorporating silhouette information and multi-level neighborhood structures to address cross-clothes invariance in unsupervised long-term re-ID, with potential impact on open-world re-ID systems and benchmarking.

Abstract

In this paper, we address a highly challenging yet critical task: unsupervised long-term person re-identification with clothes change. Existing unsupervised person re-id methods are mainly designed for short-term scenarios and usually rely on RGB cues so that fail to perceive feature patterns that are independent of the clothes. To crack this bottleneck, we propose a silhouette-driven contrastive learning (SiCL) method, which is designed to learn cross-clothes invariance by integrating both the RGB cues and the silhouette information within a contrastive learning framework. To our knowledge, this is the first tailor-made framework for unsupervised long-term clothes change \reid{}, with superior performance on six benchmark datasets. We conduct extensive experiments to evaluate our proposed SiCL compared to the state-of-the-art unsupervised person reid methods across all the representative datasets. Experimental results demonstrate that our proposed SiCL significantly outperforms other unsupervised re-id methods.
Paper Structure (19 sections, 12 equations, 5 figures, 11 tables, 1 algorithm)

This paper contains 19 sections, 12 equations, 5 figures, 11 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Long-Term Person Re-Identification
  4. Unsupervised Person Re-Identification
  5. Our Methodology: Silhouette-Driven Contrastive Learning (SiCL)
  6. Hierarchical Fusion Clustering
  7. Contrastive Learning Module
  8. Curriculum Nerighour Selecting
  9. Training and Inference Procedure for SiCL
  10. Experiments
  11. Experimental Setting
  12. Comparison to State-of-the-art Methods
  13. Ablation Study
  14. More Evaluations and Analysis
  15. Conclusion
  16. ...and 4 more sections

Figures (5)

  • Figure 1: Illustration of hierarchical fusion clustering: At a lower level, clustering is used to reveal the neighbor structure of instances. At a higher level, silhouette features and RGB features are integrated to exploit the neighbor structure of the clusters.
  • Figure 2: Visualizing the intrinsic challenges in long-term person re-id with clothes change. We randomly selected $28$ images of a single individual from the DeepChange Deepchange dataset. Evidently, the variances in visual characteristics between different outfits (rows) are considerably more pronounced compared to those within the same outfit (each column).
  • Figure 3: Our proposed Silhouette-Driven Contrastive Learning (SiCL) framework.
  • Figure 4: Visualization of the top-10 best matched images on LTCC and VC-Clothes. We show the top-10 best matching samples in the gallery set for the query sample with CACL and our proposed SiCL. The images with frames in green and in red are the correctly matched images and mismatched images, respectively.
  • Figure 5: Visualization of the top-10 best matched images on LTCC and VC-Clothes. We show the top-10 best matching samples in the gallery set for the query sample with SpCL and our proposed SiCL. The images with frames in green and in red are the correctly matched images and mismatched images, respectively.