A curriculum pseudo-labeling and consistency-regularization method lets rPPG models train on 20% labeled data plus unlabeled videos with accuracy close to full supervision.
Time–frequency learning framework for rppg signal estimation using scalogram-based feature map of facial video data,
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
-
Semi-rPPG: Semi-Supervised Remote Physiological Measurement with Curriculum Pseudo-Labeling
A curriculum pseudo-labeling and consistency-regularization method lets rPPG models train on 20% labeled data plus unlabeled videos with accuracy close to full supervision.