I have the same issue.

I have put together a command to do inference. My model has no instance or label data so I used "--no_instance" and "--label_nc 0" flags:

python test.py --name [Name of the trained model] --dataroot [path of the directory that contains input images] --label_nc 0 --no_instance --how_many 144(Number of images to do interference. If you don't give this it process 50 images) --results_dir [Directory to put results into]

Thank you for your response but i have already done this what i want is that instead of script i want to run it like api for my work by pointing towords a specific image