DSAT, a dynamic channel-gating and cross-scale attention transformer, reports small NME improvements over selected face alignment baselines on 300W, WFLW, COFW, and AFLW.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Precise Facial Landmark Detection by Dynamic Semantic Aggregation Transformer
DSAT, a dynamic channel-gating and cross-scale attention transformer, reports small NME improvements over selected face alignment baselines on 300W, WFLW, COFW, and AFLW.