Open Lucky0775 opened 2 years ago
step 1: you need to download the datasets from the Cityscapes website: https://www.cityscapes-dataset.com/
step 2: In our experiments, we download the images to 128x128 or 256x256. You need to also put images and semantic label maps in different folders (trainA
, trainB
, testA
, testB
).
step 3: Follow the CycleGAN training and test README + change the --dataroot
and --name
accordingly.
hi!Can you explain how to use the cycleGAN model to train the label (trainB) to image (trainA) task on the cityscape dataset and its training command?