AI Paper Atlas

Research map / Video / Video segmentation and tracking

rPPG-VQA: A Video Quality Assessment Framework for Unsupervised rPPG Training

Tianyang Dai, Ming Chang, Yan Chen, Yang Hu

CVPR 2026 · Video › Video segmentation and tracking · approach: Sign language and physiological video analysis · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add frequency consistency weight to simple SNR averaging (ablation)signal-level branch · PURE ablation: signal-level branchMAE: 0.82 -> 0.52Improved
add spectral correlation weight to simple SNR averaging (ablation)signal-level branch · PURE ablation: signal-level branchMAE: 0.82 -> 0.57Improved
add spectral correlation weight on top of frequency consistency weight (ablation)signal-level branch · PURE ablation: signal-level branchMAE: 0.52 -> 0.47Improved
add ranking-based filtering to WRS (ablation)TAS · PURE ablation: TAS componentsMAE: 0.71 -> 0.70No clear effect
add duration-aware sampling to WRS (ablation)TAS · PURE ablation: TAS componentsMAE: 0.71 -> 0.49Improved
combine filtering with duration-aware sampling (full TAS) (ablation)TAS · PURE ablation: TAS componentsMAE: 0.71 -> 0.47Improved
mix scene-level score into quality score (alpha 0.8 instead of 1.0) (ablation)rPPG-VQA · PURE ablation: fusion weight alphaMAE: 0.78 -> 0.47Improved
use only scene-level score (alpha 0.0 instead of 0.8) (ablation)rPPG-VQA · PURE ablation: fusion weight alphaMAE: 0.47 -> 6.91Hurt
use TAS instead of Top/Bottom K with rPPG-VQA scoreContrast-Phys · PURE test (train on curated in-the-wild data)MAE: 1.36 -> 0.47Improved
use rPPG-VQA criterion instead of IPR under WRSContrast-Phys · PURE test (train on curated in-the-wild data)MAE: 4.89 -> 0.71Improved
use rPPG-VQA criterion instead of SNR under Top/Bottom KContrast-Phys · PURE test (train on curated in-the-wild data)MAE: 2.34 -> 1.36Improved
curated Celeb-DF subset added to CAS(ME)2 instead of full Celeb-DFContrast-Phys · PURE test (train on curated in-the-wild data)MAE: 31.07 -> 0.47Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy