Disentangling 3D/4D Facial Affect Recognition With Faster Multi-View Transformer
Sign Up to like & getrecommendations! Published in 2021 at "IEEE Signal Processing Letters"
DOI: 10.1109/lsp.2021.3111576
Abstract: In this paper, we propose MiT: a novel multi-view transformer model1 for 3D/4D facial affect recognition. MiT incorporates patch and position embeddings from various patches of multi-views and uses them for learning various facial muscle… read more here.
Keywords: facial affect; affect recognition; view transformer; multi view ... See more keywords