Illumination invariants in deep video expression recognition. (April 2018)

• Record Type: Journal Article
• Authors: Gupta, Otkrist; Raviv, Dan; Raskar, Ramesh
• Is Part Of: Pattern recognition. Volume 76(2018:Apr.)
• Page Start: 25
• Page End: 35
• Publication Date: 2018-04
• Subjects: Deep learning -- Expression recognition -- Video classification -- Neural nets -- Machine learning; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2017.10.017
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract : highlights: We develop a scale invariant architecture for generating illumination invariant deep motion features. We report state of the art results for video gesture recognition using spatio-temporal convolutional neural networks. We introduce an improved topology and protocol for semi-supervised learning, where the number of labeled data points is only a fraction of the entire dataset. Abstract: In this paper we present architectures based on deep neural nets for expression recognition in videos, which are invariant to local scaling. We amalgamate autoencoder and predictor architectures using an adaptive weighting scheme coping with a reduced size labeled dataset, while enriching our models from enormous unlabeled sets. We further improve robustness to lighting conditions by introducing a new adaptive filter based on temporal local scale normalization. We provide superior results over known methods, including recent reported approaches based on neural nets.