Research map / Video / Video segmentation and tracking
Augmentation of rPPG Benchmark Datasets: Learning to Remove and Embed rPPG Signals via Double Cycle Consistent Learning from Unpaired Facial Videos
Cheng-Ju Hsieh, Wei-Hao Chung, Chiou-Ting Hsu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add double-cycle loss and rPPG loss on Embedding-Net (ablation) | RErPPG-Net · UBFC-rPPG ablation | MAE: 44.79 -> 4.14 | Improved |
| add double-cycle loss and rPPG loss on Embedding-Net (ablation) | RErPPG-Net · UBFC-rPPG ablation | PSNR: 5.11 -> 49.72 | Improved |
| add double-cycle and rPPG-removal losses on Removal-Net (ablation) | RErPPG-Net · UBFC-rPPG ablation | MAE: 4.14 -> 2.30 | Improved |
| add multi-layer perceptual loss Lp (ablation) | RErPPG-Net · UBFC-rPPG ablation | MAE: 2.30 -> 0.71 | Improved |
| use double-cycle instead of single-cycle consistency (ablation) | RErPPG-Net · UBFC-rPPG | MAE: 2.38 -> 0.71 | Improved |
| train estimator on original plus augmented videos | Ours estimator · UBFC-rPPG | MAE: 0.66 -> 0.41 | Improved |
| train rPPGNet on original plus augmented videos | rPPGNet · UBFC-rPPG | MAE: 0.72 -> 0.56 | Improved |
| train on original plus augmented videos (30/12 split of [13]) | Ours estimator · UBFC-rPPG (30/12 split of [13]) | MAE: 0.64 -> 0.75 | Hurt |
| train estimator on original plus augmented videos | Ours estimator · PURE | MAE: 0.69 -> 0.38 | Improved |
| fine-tune on augmented UBFC videos, test on PURE | Ours estimator · Cross-dataset test on PURE | MAE: 14.18 -> 4.36 | Improved |
| fine-tune on augmented PURE videos, test on UBFC | Ours estimator · Cross-dataset test on UBFC-rPPG | MAE: 3.78 -> 2.23 | Improved |
| fine-tune on Aug-rPPG, test on VIPL-HR | Ours estimator · Cross-dataset test on VIPL-HR | MAE: 28.94 -> 25.40 | Improved |
Closest papers by meaning
- Facial Video-based Remote Physiological Measurement via Self-supervised Learning (TPAMI 2023 · similarity 0.91)
- rPPG-VQA: A Video Quality Assessment Framework for Unsupervised rPPG Training (CVPR 2026 · similarity 0.91)
- Remote Photoplethysmography in Real-World and Extreme Lighting Scenarios (CVPR 2025 · similarity 0.89)
- PhysDiff: Physiology-based Dynamicity Disentangled Diffusion Model for Remote Physiological Measurement (AAAI 2025 · similarity 0.89)
- RhythmMamba: Fast, Lightweight, and Accurate Remote Physiological Measurement (AAAI 2025 · similarity 0.88)
- FactorizePhys: Matrix Factorization for Multidimensional Attention in Remote Physiological Sensing (NeurIPS 2024 · similarity 0.87)
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.
