Enhancing person re-identification via Uncertainty Feature Fusion Method and Auto-weighted Measure Combination

Person re-identification (Re-ID) is a challenging task that involvesidentifying the same person across different camera views in surveillancesystems. Current methods usually rely on features from single-camera views,which can be limiting when dealing with multiple cameras and challenges such aschanging viewpoints and occlusions. In this paper, a new approach is introducedthat enhances the capability of ReID models through the Uncertain FeatureFusion Method (UFFM) and Auto-weighted Measure Combination (AMC). UFFMgenerates multi-view features using features extracted independently frommultiple images to mitigate view bias. However, relying only on similaritybased on multi-view features is limited because these features ignore thedetails represented in single-view features. Therefore, we propose the AMCmethod to generate a more robust similarity measure by combining variousmeasures. Our method significantly improves Rank@1 accuracy and Mean AveragePrecision (mAP) when evaluated on person re-identification datasets. Combinedwith the BoT Baseline on challenging datasets, we achieve impressive results,with a 7.9% improvement in Rank@1 and a 12.1% improvement in mAP on the MSMT17dataset. On the Occluded-DukeMTMC dataset, our method increases Rank@1 by 22.0%and mAP by 18.4%. Code is available:https://github.com/chequanghuy/Enhancing-Person-Re-Identification-via-UFFM-and-AMC