文摘
We propose an unsupervised approach to person re-identification based on typical surveillance image-sequences. We present a new video representation particularly tailored for person ReID. Specifically, this representation is built up on existing action space-time features. We introduce an effective video matching algorithm, Time Shift Dynamic Time Warping (TS-DTW) and its Multi-Dimension variant MDTS-DTW, for data selective based sequence matching.