FAS-UCM generates user-specific spoof images via style transfer to train a CNN that distinguishes live and spoof faces with 0.22 average error rate on the SiW database.
Perceptual losses for real-time style transfer and super-resolution,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Style Transfer Applied to Face Liveness Detection with User-Centered Models
FAS-UCM generates user-specific spoof images via style transfer to train a CNN that distinguishes live and spoof faces with 0.22 average error rate on the SiW database.