Gait Emotion Recognition is an emerging research domain that focuses on the automatic detection of emotions from a person’s manner of walking. Deep learning-based methodologies have been proven highly effective for computer vision tasks. This paper provides a powerful deep-learning architecture for emotion recognition from gait by introducing the fusion of domain-specific discriminative features with latent deep features. The proposed Bi-Modal Deep Neural Network (BMDNN) combines salient features extracted from a deep neural network with highly-discriminating handcrafted features. The proposed architecture outperforms state-of-the-art methods in all emotional classes on the Edinburgh Locomotion MoCap Dataset.
CITATION STYLE
Bhatia, Y., Bari, A. S. M. H., & Gavrilova, M. (2022). Gait Emotion Recognition Using a Bi-modal Deep Neural Network. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 13598 LNCS, pp. 46–60). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-20713-6_4
Mendeley helps you to discover research relevant for your work.