Trained weights of Contrast-phys+

[vimal-quilt]

Hi, thanks for this wonderful work. i see the accuracy is much better then previous iteration of your method. will u update this repo demo folder with latest trained weights(0%,60% & 100%) respectively. Thanks.

[zhaodongsun]

Hi,

Thanks for your message. I will upload the trained weights later. Contrast-Phys+ can combine labeled/unsynchronized rPPG datasets and unlabeled rPPG datasets to increase training data for better performance, while previous supervised methods can only work on well-labeled rPPG datasets. In practice, If your training data is in this weakly-supervised scenario, you could try Contrast-Phys+.

[vimal-quilt]

Thanks for the information and guidance.