Empirical tests of 16 architectures on 153 subjects show camera rPPG signals contain no recoverable subject-specific pulse morphology, with all models exhibiting template collapse.
arXiv preprint arXiv:2604.00534 (2026)
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CV 3years
2026 3verdicts
UNVERDICTED 3representative citing papers
A decoupled adapter with independent spatial-temporal branches via depthwise convolutions and a dynamic augmentation strategy for long-tail data achieves first place with F1 0.43808 in a micro-gesture recognition challenge.
An end-to-end spatial-temporal transformer with illumination augmentation, Residual Temporal Standardization Module, and hybrid waveform-spectral loss achieves 0.79 bpm MAE and 0.982 correlation on a new varied-illumination dataset, outperforming PhysFormer by 93.6% in MAE.
citing papers explorer
-
Template Collapse and Information-Theoretic Limits in Camera rPPG Pulse Morphology Restoration
Empirical tests of 16 architectures on 153 subjects show camera rPPG signals contain no recoverable subject-specific pulse morphology, with all models exhibiting template collapse.
-
Spatial-Temporal Decoupled Adapter for Micro-gesture Online Recognition
A decoupled adapter with independent spatial-temporal branches via depthwise convolutions and a dynamic augmentation strategy for long-tail data achieves first place with F1 0.43808 in a micro-gesture recognition challenge.
-
Illumination-Robust Camera-Based Heart-Rate Estimation for Physiological Sensing in Robots
An end-to-end spatial-temporal transformer with illumination augmentation, Residual Temporal Standardization Module, and hybrid waveform-spectral loss achieves 0.79 bpm MAE and 0.982 correlation on a new varied-illumination dataset, outperforming PhysFormer by 93.6% in MAE.