/Retinal_Super_Resolution

Image Super Resolution Using Generative Adversarial Networks for Retinal Image Analysis

Primary LanguagePython

Retinal Super_Resolution:

Inference-Training-Testing

config.py - Configuration file with all the data paths and training/testing settings layer_def.py - definitions of all the layers SuperRes.py - SRGAN Model implementation main_SR.py - main file to run training/testing of the model

For training: Set training data path to "IMAGES" in config.py and execute "python main_SR.py" Checkpoints will be saved to the path given in "CHECKPOINT" in config.py file.

For testing: Change the "test path" in line 66 of main_SR.py to the required testing data path and execute "python main_SR.py".

If there are matching checkpoint files to the "NUM_TRAIN_EPOCHS" inside the "CHECKPOINT" path, the model will use them and run the predictions on the test data. Otherwise, it will train the model first and then do the predictions.

Citing

If you use this code, please use the following BibTeX entry.

  @inproceedings{mahapatra2017image,
  title={Image super resolution using generative adversarial networks and local saliency maps for retinal image analysis},
  author={Mahapatra, Dwarikanath and Bozorgtabar, Behzad and Hewavitharanage, Sajini and Garnavi, Rahil},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={382--390},
  year={2017},
  organization={Springer}
}