Paper Title: MetaPhys: Few-Shot Adaptation for Non-Contact Physiological Measurement (ACM CHIL-2021)
Link: https://dl.acm.org/doi/10.1145/3450439.3451870
train_higher.py is the training script of our proposed MetaPhys.
train_txt.py is the training script for regular supervised training.
train_txt_ft.py is the training script for supervised training + fine-tuning.
higher_model.py is the implemnetation of TS-CAN.
rppg_dataset.py is the dataloader for MetaPhys.
data_generator.py is the dataloader for regular supervised training.
splitter.py is the code for spliting support and query sets.
MetaPhys's dataloading schema is based on pytorch-meta (https://github.com/tristandeleu/pytorch-meta), and our inner/outer loop optimizaton is based on higher (https://github.com/facebookresearch/higher).