Multiclass Semantic segmentation in pytorch

Greetings everyone

I am new to using PyTorch. I am trying to implement semantic segmentation for identifying multi-class objects for eg if I pass an image with 2 laptops and one knife, I want my model to categorise both the laptops in one class, the knife in another and background being the last class. I have understood the concepts intuitively and was able to predict binary classification using DeepLab V3. But after several tries I am failing to do the same when it comes to multi-channels. I have customised my images and masks using Gimp. I need help with the code if someone could send me a running code or a repository it’ll be a huge favour.