Research map / 3D vision / Human pose estimation
egoPPG: Heart Rate Estimation from Eye-Tracking Cameras in Egocentric Systems to Benefit Downstream Vision Tasks
Björn Braun, Rayan Armani, Manuel Meier, Max Moebus, Christian Holz
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 motion-informed temporal attention (MITA) using IMU (ablation) | PulseFormer · egoPPG-DB (5-fold, 60s windows) | HR MAE (bpm): 8.82 -> 7.67 | Improved |
| add spatial attention (SA) modules (ablation) | PulseFormer · egoPPG-DB (5-fold, 60s windows) | HR MAE (bpm): 10.49 -> 7.67 | Improved |
| add MITA, dancing activity (ablation) | PulseFormer · egoPPG-DB (5-fold, 60s windows), activity: Dancing | HR MAE (bpm): 10.54 -> 7.85 | Improved |
| add MITA, walking activity (ablation) | PulseFormer · egoPPG-DB (5-fold, 60s windows), activity: Walking | HR MAE (bpm): 8.29 -> 8.23 | No clear effect |
| use skin region instead of eye region as input (ablation) | signal-processing baseline · egoPPG-DB (5-fold, 60s windows) | HR MAE (bpm): 14.60 -> 12.40 | Improved |
| use eye tracking video at 10 fps instead of 30 fps (ablation) | PulseFormer · egoPPG-DB (frame rate study) | HR MAE (bpm): 7.67 -> 11.13 | Hurt |
| upsample 10 fps video to 30 fps by interpolation (ablation) | PulseFormer · egoPPG-DB (frame rate study) | HR MAE (bpm): 11.13 -> 10.18 | Improved |
| replace FactorizePhys with PulseFormer architecture | rPPG HR estimation · egoPPG-DB (5-fold, 60s windows) | HR MAE (bpm): 10.07 -> 7.67 | Improved |
| add predicted HR features to egocentric video | TimeSFormer · EgoExo4D proficiency estimation (overall) | top-1 accuracy (%): 39.69 -> 45.29 | Improved |
| add predicted HR features to exocentric video | TimeSFormer · EgoExo4D proficiency estimation (overall) | top-1 accuracy (%): 34.75 -> 37.67 | Improved |
| add predicted HR features to ego+exo video | TimeSFormer · EgoExo4D proficiency estimation (overall) | top-1 accuracy (%): 39.00 -> 43.94 | Improved |
| add predicted HR, cooking egocentric | TimeSFormer · EgoExo4D proficiency estimation (Cooking, Ego) | top-1 accuracy (%): 20.00 -> 40.00 | Improved |
Closest papers by meaning
- egoEMOTION: Egocentric Vision and Physiological Signals for Emotion and Personality Recognition in Real-world Tasks (NeurIPS 2025 · similarity 0.88)
- WildPPG: A Real-World PPG Dataset of Long Continuous Recordings (NeurIPS 2024 · similarity 0.88)
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.86)
- EgoLife: Towards Egocentric Life Assistant (CVPR 2025 · similarity 0.85)
- RhythmMamba: Fast, Lightweight, and Accurate Remote Physiological Measurement (AAAI 2025 · similarity 0.85)
- Ego4o: Egocentric Human Motion Capture and Understanding from Multi-Modal Input (CVPR 2025 · similarity 0.84)
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.
