train on my data

[hadi1994shokati]

Hello. Thanks for sharing the code.
I want to train the model with my data. Instead of 20 Pascal classes, my data has only 1 class and it is binary classification. Should I just use a split? What parts of the code should I change?
Looking forward to hearing you.

[tianzhuotao]

In this case, you may change the dataset.py and let the training/val use the same split, as there is only a single class in your dataset.

[hadi1994shokati]

Thank you so much for your answer. I want to use this code for soil erosion binary segmentation. My segmentation task doesn't have base classes and just has 0 and 1 (erosion and non-erosion). Is this code able to do this segmentation? If yes, Should I just change the getitem function in the dataset file? I tried to use 1 split for my data but I got an assertion error in line 171 of the dataset file which means label_class is 0 in some cases.
Thank you in advance.