Skip to content

Latest commit

 

History

History
46 lines (37 loc) · 1.97 KB

README.md

File metadata and controls

46 lines (37 loc) · 1.97 KB

Anomaly-Aware-rPPG

(Official repo for BIOSIGNALS 2023 paper)

Official BIOSIGNALS 2023 Paper: Anomaly-Aware Remote Photoplethysmography

Link to arXiv Paper: Anomaly-Aware Remote Photoplethysmography

Preprocessing

First check the preprocessing code in src/preprocessing to prepare all of the video datasets you have available.

A list of the datasets used in the paper are:

Training

Models can be trained by using the script, which has preset parameters which can be adjusted for the experiment:

cd scripts
sh train.sh  # for normal training
sh train_negative.sh  # for training with negative samples

Testing after Model Training

To make predictions on the testing datasets with your different models, you can use the validate_emitter.py script:

cd src
python validate_emitter.py

Feature Extraction and SVM fitting to Predict Anomalies

All of the feature extraction and SVM fitting code is in the testing folder. See the order of processing in the testing/README.md file. In general, you can run the python scripts by running the bash scripts in the testing/scripts folder. For example:

cd testing/scripts
## Edit the predict.sh file to specify the model and dataset you want to use
sh predict.sh