### How to Train Models Using UBFC-rPPG or PURE Datasets?

[alaahejer]

Hello,

I am trying to train one of the rPPG models (e.g., DeepPhys, BigSmall) using the UBFC-rPPG or PURE dataset. However, I couldn't find detailed instructions on how to correctly set up the dataset and execute the training script.

Could you please provide guidance on:

1. The correct directory structure for the datasets?
2. The exact command to train a model using one of these datasets?
3. Any necessary modifications to configuration files?

I appreciate your help!

Best regards,
Alaa Hajr

[alaahejer]

hello,

[SpicyYeol]

When you run main.py in the rppg directory, the training process will begin.

The reference configuration file used during training is rppg/configs/base_config.yaml.

To use the dataset, simply extract it as it was originally provided and place it under the dataset_root_path directory with its respective dataset name.

For example, if data_root_path is set to "/ssd/ssd0/dataset/raw/rppg/", then datasets should be placed as follows:

• "/ssd/ssd0/dataset/raw/rppg/PURE"
• "/ssd/ssd0/dataset/raw/rppg/UBFC"

This ensures the dataset is correctly recognized during training.