wnet-unsupervised-image-segmentation icon indicating copy to clipboard operation
wnet-unsupervised-image-segmentation copied to clipboard

Training on customized dataset

Open soumya-ranjan-sahoo opened this issue 5 years ago • 1 comments

How do I train the model using my own datasets? If you can help me with the commands please and the changes I need to make.

soumya-ranjan-sahoo avatar Sep 24 '20 11:09 soumya-ranjan-sahoo

Hi there! I know it's a bit late, but I just adapted the code to my data. In case you or someone else still needs it, all you need to do is store your train images in two folders: one for the original images and one for the masks (binary for me bc I only have 2 classes). In my case, image and corresponding mask have the same filename, so all I needed to do was adjust the get_matching_file_names function in format_data, as well as the input_size. Then you should be able to run the rest of format_data.py and create a custom .hdf5-file

JaninaPa avatar Apr 12 '21 16:04 JaninaPa