pytorch-CycleGAN-and-pix2pix
pytorch-CycleGAN-and-pix2pix copied to clipboard
junyanz
Is it normal to use 2 black and white images?
During the training with the colorize mode, something caught my attention and that is that during the training it uses two black and white images when one is in color.
This should be the norm, right?
But this....?
Check what your input_nc and output_nc values are. 1 is for grayscale, 3 is for RGB.
is that I do not put those arguments
Check what your input_nc and output_nc values are. 1 is for grayscale, 3 is for RGB.
is that I do not put those arguments
Check what your input_nc and output_nc values are. 1 is for grayscale, 3 is for RGB.
Check if you put those arguments in your training configuration. If your input images are grayscale, use input_nc=1. If the images are colored, use input_nc=3
is that I do not put those arguments
Check what your input_nc and output_nc values are. 1 is for grayscale, 3 is for RGB.
Check if you put those arguments in your training configuration. If your input images are grayscale, use input_nc=1. If the images are colored, use input_nc=3
no, I didn't use them, I used the ones that came with the colab and with the one to resume training, but nothing else.
or maybe they have defaults?
It appears that for colorization dataset, the defaults are input_nc 1 and output_nc 2. So the issue may be somewhere else. T Checking the code and data you use for training would be useful.
or maybe they have defaults?
It appears that for colorization dataset, the defaults are input_nc 1 and output_nc 2. So the issue may be somewhere else. T Checking the code and data you use for training would be useful.
If I change the output to 3 should it work?