Parallelized_LSM_for_Unintentional_Action_Recognition

• To train and validate the PLSM model, downlaod this repository.

• Download the Oops dataset and place the train and val folders inside PATH/TO/datasets/

• Make sure the paths to the dataset and annotations match the ones mentioned inside basic_pipeline_v4.0.py

• To downlaod the specific LSM model used in our work, download the model from https://drive.google.com/file/d/1F79GB-2PKKffvG2Ij5ylSUTvm5lQRUci/view?usp=sharing and transfer the model to location saved_models/

• Make sure you have created two empty folders named 'train' and 'val' inside MRI_data/PLSM_readout_100timesteps/

• Run the basic_pipeline_v4.0.py script to generate data using the LSM.

• Make sure the datapaths to the generated data matches the one used in the G_MRI_LR.py script.

• Once the data is generated, run the script G_MRI_LR.py for training the 3D CNN readout layer.