CanonicalPhys: Pose-Robust Remote Photoplethysmography via Canonical-Space Priors
Ranking
Overall
79
Content
85
Popularity
65
Observed public metrics from 1 member.
Merged summary
TL;DR - CanonicalPhys improves camera-based heart-rate estimation under head rotation by warping facial landmarks into a canonical coordinate space. This enables anatomy-dependent physiological priors and reduces pose-related errors without adding trainable parameters.
- Uses a differentiable four-point homography to stabilize facial anatomy across poses.
- Adds Lambertian pixel weighting, cross-region temporal consistency, and POS-based knowledge distillation.
- Reduces frontal-to-large-yaw MAE degradation on MMPD from 1.60Ă— to 1.33Ă—.
- Achieves cross-dataset MAE reductions of up to 32% on pose-rich targets.
Sources (1)
CanonicalPhys: Pose-Robust Remote Photoplethysmography via Canonical-Space Priors
Public signals
Hugging Face upvotes 1
TL;DR - CanonicalPhys improves camera-based heart-rate estimation under head rotation by warping facial landmarks into a canonical coordinate space. This enables anatomy-dependent physiological priors and reduces pose-related errors without adding trainable parameters.
- Uses a differentiable four-point homography to stabilize facial anatomy across poses.
- Adds Lambertian pixel weighting, cross-region temporal consistency, and POS-based knowledge distillation.
- Reduces frontal-to-large-yaw MAE degradation on MMPD from 1.60Ă— to 1.33Ă—.
- Achieves cross-dataset MAE reductions of up to 32% on pose-rich targets.