ORFormer uses per-patch messenger tokens to detect and recover occluded face regions, reducing landmark error on WFLW and COFW.
Training deep networks for facial ex- pression recognition with crowd-sourced label distribution
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
ORFormer: Occlusion-Robust Transformer for Accurate Facial Landmark Detection
ORFormer uses per-patch messenger tokens to detect and recover occluded face regions, reducing landmark error on WFLW and COFW.