This project is reimplementation of the Jakeret's U-Net.
I reimplemented the code in my own style and modified it that can train multiple patch sizes at once.
# Crop image
python "ImageCropping.py"
# Train U-Net
python "Training.py"
# Run TensorBoard
tensorboard --logdir="$TRAINING_LOG_PATH"
# Run demo
python "Demo.py"
You can change the settings using config.ini and Setting.py.