How to train on my dataset?

Thanks for your share!
I want to konw how to train on my datasets?
My image and label is RGB

@zxshi Currently the code accepts image in .jpg and labels in .png format. In order to convert labels from RGB to .png format please follow encode_segmap

sunets/ptsemseg/loader/pascal_voc_loader.py

Line 164 in 11057da

def encode_segmap(self, mask):

and get_pascal_labels

sunets/ptsemseg/loader/pascal_voc_loader.py

Line 158 in 11057da

def get_pascal_labels(self):

functions. Note that this converts RGB labels to index. Index can be in any order as far as it is kept same across train and test set. In get_pascal_labels, replace pascal dataset labels with labels from your dataset.

OK, thank you very much!
I will try it.

I noticed that you have used the network to get good results in the land cover classification challenge. Can you share the complete code in this challenge?