UNet Training for MRI Image Segmentation
- The code is the slight modification of this implementation
- You can use Inputs and Labels folders to upload your own input and label images. Input images are (256,256,3) RGB jpeg images. Output labels are binary (256,256) png images.
- Label images are generated by using this repo which utilizes the level-set method.
- There are 10 standard and 10 augmented samples for you to try.
- You can load the pre-trained model parameters from this file
- You can use this jupyter notebook to load trained model and perform predictions.
- You can use this jupyter notebook to train your own model with your training input-label pairs.