Self-supervised on-line cumulative learning from video streams

We present a novel online self-supervised method for face identity learning from video streams. The method exploits deep face feature descriptors together with a memory based learning mechanism that takes advantage of the temporal coherence of visual data. Specifically, we introduce a discriminative descriptor matching solution based on Reverse Nearest Neighbour and a memory based cumulative learning strategy that discards redundant descriptors while time progresses. This allows building a comprehensive and cumulative representation of all the past visual information observed so far. It is shown that the proposed learning procedure is asymptotically stable and can be effectively used in relevant applications like multiple face identification and tracking from unconstrained video streams. Experimental results show that the proposed method achieves comparable results in the task of multiple face tracking and better performance in face identification with offline approaches exploiting future information.