ChatPPG.
Join free
Model

PHA + FDP

progressive hierarchical attention and frequency-decoupled loss for unsupervised rPPG

RGBPhysiological Measurement
Cite
Venue listedCatalog paper date: 2026Source
CitationsView DOI record (opens in a new tab)

Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.

Single-branch 3D CNN trained with Contrast-Phys-style PSD contrastive learning. PHA cascades spatial, temporal, and channel attention with learnable stage fusion. FDP loss anchors each waveform at a dynamic fundamental frequency, suppresses out-of-band power, and applies confidence-gated harmonic constraints, so training matches the band-pass plus dominant-peak HR readout used at test time.

Signals
RGB
Form factor
camera_rppg
License
Paper under IOP standard licence
Year
2026
Architecture

3D CNN with progressive hierarchical attention and frequency-decoupled physiological loss

At a glance
Type
Task model
Modalities
RGB
Form factor
Camera rPPG
Parameters
Not reported
Architecture
3D CNN with progressive hierarchical attention and frequency-decoupled physiological loss
Venue
Physiological Measurement (IOP), published 2026-10-06, DOI 10.1088/1361-6579/aeac9a
Year
2026
arXiv
Not reported
DOI
10.1088/1361-6579/aeac9a (opens in a new tab)
Code
Not reported
Weights
Not reported
Licence
Paper under IOP standard licence
Caveats
Not reported
Datasets
Not reported
Added
Not reported
Updated
Oct 27, 2026
PHA + FDP: ChatPPG research summary card

How it works

How PHA + FDP works: Input signals → Model → Downstream tasksInput signals: RGB. Model: PHA + FDP (Transformer · Convolutional network).Input signalsRGBRGBModel×NPHA + FDPTransformer · Convolutional network
Architecture
3D CNN with progressive hierarchical attention and frequency-decoupled physiological loss
Inputs
RGB
Approach
rppg · model · unsupervised · contrastive · attention · spectral loss

Datasets

No public datasets linked yet.

Cite

APA

PHA + FDP. (2026). Physiological Measurement. https://doi.org/10.1088/1361-6579/aeac9a

BibTeX

@article{phafdpunsupervis2026,
  title         = {{PHA + FDP}},
  year          = {2026},
  journal       = {Physiological Measurement},
  doi           = {10.1088/1361-6579/aeac9a}
}

Built from the hub record. Check it against the publisher before submitting.

History

  1. Record updated

Same tasks, data and authors