Incremental Learning from Low-labelled Stream Data in Open-Set Video Face Recognition. (November 2022)

• Record Type: Journal Article
• Authors: Lopez-Lopez, Eric; Pardo, Xose M.; Regueiro, Carlos V.
• Is Part Of: Pattern recognition. Volume 131(2022)
• Publication Date: 2022-11
• Subjects: Open-set face recognition -- Incremental Learning -- Self-updating -- Adaptive biometrics -- Video-surveillance; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2022.108885
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A online approach to unsupervised instance-incremental learning with stream data. Adaptation from pseudo-labels, which are the own predictions of the system. A strategy to deal with catastrophic forgetting and the effect of wrong pseudo-labels. Designed to operate in the open-set, extendable to the class-incremental problem. Method for person re-identification based on face without a reservoir of face images. Abstract: Deep Learning approaches have brought solutions, with impressive performance, to general classification problems where wealthy of annotated data are provided for training. In contrast, less progress has been made in continual learning of a set of non-stationary classes, mainly when applied to unsupervised problems with streaming data. Here, we propose a novel incremental learning approach which combines a deep features encoder with an Open-Set Dynamic Ensembles of SVM, to tackle the problem of identifying individuals of interest (IoI) from streaming face data. From a simple weak classifier trained on a few video-frames, our method can use unsupervised operational data to enhance recognition. Our approach adapts to new patterns avoiding catastrophic forgetting and partially heals itself from miss-adaptation. Besides, to better comply with real world conditions, the system was designed to operate in an open-set setting. Results show a benefit of up to 15% F1-score increase respect to non-adaptive state-of-the-art methods.