I want to convert RGB street images to a 2-class semantic label. I have prepared the dataset accordingly. While training, I added the arguments "--input_nc 2 --label_nc 3" but that's not how the expected encoding is, it seems. What changes do I need to make to my custom data loader and vid2vid model to train this?
Edit: I suppose variables _transformI and _transformL in _data/fewshot_customdataset.py need to be changed to something else, but I'm not sure to what.
ChintanTrivedi
commented
3 years ago
I want to convert RGB street images to a 2-class semantic label. I have prepared the dataset accordingly. While training, I added the arguments "--input_nc 2 --label_nc 3" but that's not how the expected encoding is, it seems. What changes do I need to make to my custom data loader and vid2vid model to train this?
Edit: I suppose variables _transformI and _transformL in _data/fewshot_customdataset.py need to be changed to something else, but I'm not sure to what.