A 3D-VQGAN with spatial-temporal codebooks and code-lookup transformers restores compressed face videos and removes flicker in about 3 seconds per 24-frame clip.
Stabilization of flicker- like effects in image sequences through local contrast cor- rection
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
-
Efficient Video Face Enhancement with Enhanced Spatial-Temporal Consistency
A 3D-VQGAN with spatial-temporal codebooks and code-lookup transformers restores compressed face videos and removes flicker in about 3 seconds per 24-frame clip.