This academic article discusses emotion estimation in human-robot interaction (HRI). The authors present a new dataset, HRI-A VC, obtained from a human-robot collaborative task. The data set is unique because it includes self-reported arousal and valence values directly from humans, instead of relying on expert annotations. The authors also propose a spatial and temporal attention-based network to estimate emotions from these image frames. The study suggests that an attention-based network can successfully estimate valence and arousal from the data set, even when these values are not available per frame.
Publication date: 23 Oct 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2310.12887
