Abstract
Person re-identification (ReID) plays an important role in intelligent surveillance and receives widespread attention from academics and the industry. Due to extreme changes in viewing angles, some discriminative local regions are suppressed. In addition, the data with similar backgrounds collected by a fixed viewing angle camera will also affect the model’s ability to distinguish a person. Therefore, we need to discover more fine-grained information to form the overall characteristics of each identity. The proposed self-erasing network structure composed of three branches benefits the extraction of global information, the suppression of background noise and the mining of local information. The two self-erasing strategies that we proposed encourage the network to focus on foreground information and strengthen the model’s ability to encode weak features so as to form more effective and richer visual cues of a person. Extensive experiments show that the proposed method is competitive with the advanced methods and achieves state-of-the-art performance on DukeMTMC-ReID and CUHK-03(D) datasets. Furthermore, it can be seen from the activation map that the proposed method is beneficial to spread the attention to the whole body. Both metrics and the activation map validate the effectiveness of our proposed method.
Highlights
Person re-identification is a task that uses computer vision technology to retrieve the identity of a specific person in surveillance with multiple non-overlapping cameras
The contributions of this work can be summarized as follows: (1) We designed an end‐to‐end multi‐branch self‐erasing network to ov ground interference and extract sufficient local information; The contributions of this work can be summarized as follows: (1) We designed an end-to-end multi-branch self-erasing network to overcome background interference and extract sufficient local information; (2) Through the visualization of activation maps, we found that fully digging out a more comprehensive feature representation of a person helps to improve the discriminability of the model
We used cumulative match characteristic (CMC) and the mean average precision metrics to evaluate the quality of person re-identification models
Summary
Person re-identification is a task that uses computer vision technology to retrieve the identity of a specific person in surveillance with multiple non-overlapping cameras. The performance of ReID continues to improve with the help of deep convolutional neural networks, it still faces many challenges, e.g., various background clutter, light changes, occlusions and changes in attitude, which will affect the model feature coding. The ReID model needs to find richer discriminative visual cues to form the comprehensive characteristics of each identity. Convolutional Baseline (PCB) [9] uses a refined part pooling strategy to ma closer to the real region; (3) Optimization strategy—some works a3totfe17mpted ter person features with the modification of the backbone architecture, data and regularization. (2) Through the visualization of activation maps, we found that fully digging out a more comprehensive feature representation of a person helps to improve the discriminability of the model. (3) Extensive experiments show that the proposed model has certain advantages over recent state-of-the-art methods
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
