Multi-frame techniques for long-term people re-identification with consumer depth cameras

In this work we address the people re-identification problem with particular attention to long-term re-identification in which a subject has to be re-identified even after some days from the last occurrence. In particular, we focus on multi-frame techniques, which exploit information from multiple frames for producing the re-identification output. We introduced three real-time algorithms which improve classification performance obtained by state-of-the-art algorithms on two public datasets.