Multi-branch Person Re-identification Net

Person ReID is a complex and important problem within the domain of computer vision. Its objective is to identify and match individuals in video sequences, often captured from different camera viewpoints. By comparing visual characteristics, poses, and clothing of pedestrians, Person ReID enables the tracking and recognition of the same person across various scenarios. This paper introduces a Multi-Branches Person Re-identification Net (MPRN), a novel framework for person re-identification. MPRN leverages OSNet as the backbone and establishes a multi-branch structure on top of it, effectively integrating both local and global features. Experimental results demonstrate that the combination of OSNet and the multi-branch structure yields highly competitive results. Specifically, it achieves a mean average precision (mAP) of 90.5% and a rank-1 accuracy of 96% on the Market-1501 dataset, as well as an mAP of 85.5% and a rank-1 accuracy of 87.7% on the CHUK03-L dataset.