OC4-ReID: Occluded Cloth-Changing Person Re-Identification

TL;DR

This paper introduces OC4-ReID, a new challenging task for person re-identification that involves clothing changes and occlusions, along with new datasets, a benchmark, and a novel learning method.

Contribution

It proposes the OC4-ReID task, creates new occluded datasets, and develops a benchmark with a new screening module and loss function for partial feature learning.

Findings

Outperforms existing methods on new occluded datasets

Demonstrates robustness to occlusion and clothing changes

Provides publicly available datasets and code

Abstract

The study of Cloth-Changing Person Re-identification (CC-ReID) focuses on retrieving specific pedestrians when their clothing has changed, typically under the assumption that the entire pedestrian images are visible. Pedestrian images in real-world scenarios, however, are often partially obscured by obstacles, presenting a significant challenge to existing CC-ReID systems. In this paper, we introduce a more challenging task termed Occluded Cloth-Changing Person Re-Identification (OC4-ReID), which simultaneously addresses two challenges of clothing changes and occlusion. Concretely, we construct two new datasets, Occ-LTCC and Occ-PRCC, based on original CC-ReID datasets to include random occlusions of key pedestrians components (e.g., head, torso). Moreover, a novel benchmark is proposed for OC4-ReID incorporating a Train-Test Micro Granularity Screening (T2MGS) module to mitigate the influence of occlusion and proposing a Part-Robust Triplet (PRT) loss for partial features learning. Comprehensive experiments on the proposed datasets, as well as on two CC-ReID benchmark datasets demonstrate the superior performance of proposed method against other state-of-the-art methods. The codes and datasets are available at: https://github.com/1024AILab/OC4-ReID.