CorrDetail achieves state-of-the-art face forgery detection by training a vision-language model to self-correct error-laden visual questions and combining it with a dedicated visual branch.
An image is worth 16x16 words: Transformers for image recognition at scale
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
-
CorrDetail: Visual Detail Enhanced Self-Correction for Face Forgery Detection
CorrDetail achieves state-of-the-art face forgery detection by training a vision-language model to self-correct error-laden visual questions and combining it with a dedicated visual branch.