Project link: https://github.com/ShuangLI59/person_search

This project is the implementation of the paper by Tong Xiao, and others that focuses on the problem of person detection and re-identification. This project can be used in video surveillance. The existing person re-identification methods mainly assume that the person is cropped and aligned. However, in real-world scenarios, the person detection algorithm may fail to extract the perfect crop region of the person and lower the identification accuracy. In this project, the authors solve detection and identification jointly in a novel architecture inspired by Faster RCNN. The current project is implemented in the Caffe Deep Learning Framework.