Unsupervised Incremental Learning of Deep Descriptors From Video Streams

TL;DR

This paper introduces an unsupervised incremental learning method for face identity recognition from video streams, leveraging deep neural networks and a novel feature matching strategy to enable stable, memory-controlled learning over time.

Contribution

It proposes a new unsupervised learning approach that uses temporal coherence and a reverse nearest neighbor matching for incremental face recognition from videos.

Findings

The method is asymptotically stable.

It effectively supports multiple face tracking.

Memory size is controlled during learning.

Abstract

We present a novel unsupervised method for face identity learning from video sequences. The method exploits the ResNet deep network for face detection and VGGface fc7 face descriptors together with a smart learning mechanism that exploits the temporal coherence of visual data in video streams. We present a novel feature matching solution based on Reverse Nearest Neighbour and a feature forgetting strategy that supports incremental learning with memory size control, while time progresses. It is shown that the proposed learning procedure is asymptotically stable and can be effectively applied to relevant applications like multiple face tracking.