Search papers, labs, and topics across Lattice.
2
0
3
TTSD-FAR maintains consistent performance in emotion recognition even when up to half of the input modalities are missing, outperforming traditional adaptation methods.
Despite advances in multimodal deep learning, recognizing subtle emotional states like ambivalence and hesitancy from video remains a significant challenge, even for state-of-the-art models.
