Memory Based Online Learning of Deep Representations from Video Streams

Abstract

We present a novel online unsupervised method for face identity learning from video streams. The method exploits deep face descriptors together with a memory based learning mechanism that takes advantage of the temporal coherence of visual data. Specifically, we introduce a discriminative feature matching solution based on Reverse Nearest Neighbour and a feature forgetting strategy that detect redundant features and discard them appropriately while time progresses. 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. Code will be publicly available.

arXiv admin note: text overlap with arXiv:1708.03615

References in corpus (6)

• Progressive Neural Networks
• MOTChallenge 2015: Towards a Benchmark for Multi-Target Tracking
• Revisiting Unreasonable Effectiveness of Data in Deep Learning Era
• Learning to Remember Rare Events
• Simple Online and Realtime Tracking with a Deep Association Metric
• Tracking the Trackers: An Analysis of the State of the Art in Multiple Object Tracking

Cited by in corpus (9)

• Deep Learning in Video Multi-Object Tracking: A Survey
• Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIP
• Class-incremental Learning with Pre-allocated Fixed Classifiers
• Regular Polytope Networks
• CL2R: Compatible Lifelong Learning Representations
• Siamese Natural Language Tracker: Tracking by Natural Language Descriptions with Siamese Trackers
• CoReS: Compatible Representations via Stationarity
• Towards Open World Object Detection
• CAN: Composite Appearance Network for Person Tracking and How to Model Errors in a Tracking System