A facial expression recognition model that fuses dense segmentation priors and sparse landmark priors, plus a new dynamic margin loss and a new occlusion dataset, reports state-of-the-art accuracy on RAF-DB and AffectNet.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Rethinking Occlusion in FER: A Semantic-Aware Perspective and Go Beyond
A facial expression recognition model that fuses dense segmentation priors and sparse landmark priors, plus a new dynamic margin loss and a new occlusion dataset, reports state-of-the-art accuracy on RAF-DB and AffectNet.