Incremental Learning from Low-labelled Stream Data in Open-Set Video Face Recognition

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 a set non-stationary classes, mainly when applied unsupervised streaming data. Here, we propose novel incremental approach which combines deep features encoder an Open-Set Dynamic Ensembles SVM, tackle the problem identifying individuals interest (IoI) from face From simple weak classifier trained on few video-frames, our method can use operational enhance recognition. Our adapts new patterns avoiding catastrophic forgetting and partially heals itself miss-adaptation. Besides, better comply real world conditions, system was designed operate open-set setting. Results show benefit up 15% F1-score increase respect non-adaptive state-of-the-art methods.