Explainable graph-attention based person re-identification in outdoor conditions

Person re-identification is the process of recognizing an individual across multiple camera views. It is essential for an extensive range of applications related to security and biometrics. We propose a shift in perspective for the ongoing re-identification studies. Present graph-based person re-identification methods need to explain the importance of graph attention and convolution techniques. However, our proposed method focuses on a less intrusive and explainable approach to attention selection and graph convolution methods. The proposed multi-channel framework utilizes visual features and attribute labels to represent each person uniquely. We applied large-scale benchmark datasets, such as MSMT17, DukeMTMC, CUHK03, and Market-1501.