This is the implementation of the PI-DDPM network using a basic UNet architecture as backbone.
- To train the model place your data generated by the dataset_generation script (if you are generating simulated data) or the STORM script if you are generating the respective STORM dataset.
- In the train_ddpm or train_unet script change the paths of the loading data to the ones that you generated.
- Choose a training modality, either widefield or confocal.
- Run the script.
- To test the model generate your testing dataset using the dataset_generation script.
- Change the paths corresponding to your data.
- Change the paths to the weights files that you wish to use.
- Run the test script.