Model

Contrast-Phys / Contrast-Phys+

unsupervised contrastive rPPG

UnconfirmedCatalog paper date: Not reported

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.

Contrast-Phys is an unsupervised contrastive learning method for remote PPG from face video (ECCV 2022), with Contrast-Phys+ follow-on work. Official MIT-licensed code is at github.com/zhaodongsun/contrast-phys.

Form factor
camera_rppg
License
MIT
Openness
Open
Code
github.com/zhaodongsun/contrast-phys
Published
ECCV 2022
At a glance
Type
Task model
Modalities
Not reported
Form factor
Camera rPPG
Parameters
Not reported
Architecture
Not reported
Venue
Not reported
Year
Not reported
arXiv
Not reported
DOI
Not reported
Code
Not reported
Weights
Not reported
Caveats
Not reported
Datasets
Not reported
Added
Not reported
Updated
Oct 24, 2026

Original description

Update: The paper for Contrast-Phys+ has been accepted by TPAMI. Please check the contrast-phys+ folder for the code.

We incorporate prior knowledge about remote photoplethysmography (rPPG) into contrastive learning to achieve unsupervised rPPG training. The method does not require any ground truth for training.

Please check requirement.txt for the required Python libraries.

Verbatim fromGitHub· Description from the repository README; licence: MIT

Contrast-Phys / Contrast-Phys+: ChatPPG research summary card

How it works

Inputs
Contrast-Phys / Contrast-Phys+
Outputs / tasks
Downstream tasks
Approach
rppg · model · contrast phys · unsupervised · mit

Datasets

No public datasets linked yet.

Cite

APA

Contrast-Phys / Contrast-Phys+. (n.d.).

BibTeX

@misc{contrastphyscontrastphys,
  title         = {{Contrast-Phys / Contrast-Phys+}}
}

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

History

  1. Record updated