Same-clothes person re-identification with dual-stream network

Zhiyue Wu,Zirui Hu,Jianwei Ding
DOI: https://doi.org/10.1007/s00530-024-01269-0
IF: 3.9
2024-02-27
Multimedia Systems
Abstract:Person re-identification (Re-ID) has long been a pressing challenge in the field of computer vision, with researchers primarily focusing on issues such as occlusion, clothing changes, and cross-modality scenarios. However, there has been a lack of research specifically addressing the re-identification of pedestrians wearing identical clothing. In this study, we explore and investigate this unique scenario for the first time, assuming that all detected pedestrians are dressed in the same attire. To evaluate the effectiveness of existing person re-identification methods, we establish a validation dataset comprising synthetic data. Additionally, we propose a novel dual-stream feature learning framework model to address the issue of clothing similarity in person re-identification. Our experimental results demonstrate that our model not only tackles the same-clothing challenge but also exhibits strong adaptability and robustness in cloth-changing Re-ID tasks. We believe that this research will encourage further attention from researchers toward the same-clothing problem in person re-identification, and the associated code and dataset are made available on https://github.com/Titor99/SC-ReID.
computer science, information systems, theory & methods
What problem does this paper attempt to address?