Hi, how to deal with one-channel(i mean gray scale) images with SRGAN's network?

[leftthomas]

Hi, how to deal with one-channel(i mean gray scale) images with SRGAN's network?

[leftthomas]

@CaptainEven change the in channel of the network to 1, and other misc

[CaptainEven]

Hi, i changed input channels from 3 to 1 in Generator and Discriminator but an error occured like this: File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\container.py", line 67, in forward
input = module(input)
File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 325, in call
result = self.forward(*input, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\conv.py", line 277, in forward
self.padding, self.dilation, self.groups)
File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\functional.py", line 90, in conv2d
return f(input, weight, bias)
RuntimeError: Given groups=1, weight[64, 3, 3, 3], so expected input[8, 1, 88, 88] to have 3 channels, but got 1 channels instead
got any idea or suggestion?

[leftthomas]

@CaptainEven Because you only change that, you should also change other misc, not just only there, you should change the image preprocessing code, visualization code, some save image code and PSNR computation code, etc.

[apyrros]

In data_utils.py simply add .convert('RGB') to convert a grayscale image to RGB and add three channels. It is a built-in function in PIL.

class TrainDatasetFromFolder(Dataset):
def init(self, dataset_dir, crop_size, upscale_factor):
super(TrainDatasetFromFolder, self).init()
self.image_filenames = [join(dataset_dir, x) for x in listdir(dataset_dir) if is_image_file(x)]
crop_size = calculate_valid_crop_size(crop_size, upscale_factor)
self.hr_transform = train_hr_transform(crop_size)
self.lr_transform = train_lr_transform(crop_size, upscale_factor)

def __getitem__(self, index):
    hr_image = self.hr_transform(Image.open(self.image_filenames[index]).convert('RGB'))
    lr_image = self.lr_transform(hr_image)
    return lr_image, hr_image

def __len__(self):
    return len(self.image_filenames)

class ValDatasetFromFolder(Dataset):
def init(self, dataset_dir, upscale_factor):
super(ValDatasetFromFolder, self).init()
self.upscale_factor = upscale_factor
self.image_filenames = [join(dataset_dir, x) for x in listdir(dataset_dir) if is_image_file(x)]

def __getitem__(self, index):
    hr_image = Image.open(self.image_filenames[index]).convert('RGB')
    w, h = hr_image.size
    crop_size = calculate_valid_crop_size(min(w, h), self.upscale_factor)
    lr_scale = Resize(crop_size // self.upscale_factor, interpolation=Image.BICUBIC)
    hr_scale = Resize(crop_size, interpolation=Image.BICUBIC)
    hr_image = CenterCrop(crop_size)(hr_image)
    lr_image = lr_scale(hr_image)
    hr_restore_img = hr_scale(lr_image)
    return ToTensor()(lr_image), ToTensor()(hr_restore_img), ToTensor()(hr_image)