A simple U-Net implementation for biomedical image segmentation.
To run the U-Net:
- Create a folder data in the same directory as other files.
- Create folders npydata, results, train and test inside data folder.
- Create folders images and labels inside both the train and test folders. All images must of type jpg.
- Place your training images and their labels(mask) inside
./data/train/images
and./data/train/labels
and place your testing images under./data/test/images
. Make sure that all the resolution of your images are a multiple of 32. Like 640x960 or 512x512. - Run python data.py
- Run python unet.py and wait for the training to happen. Once complete, your results will be placed under
./data/results
.