yiyang7 / super_resolution_with_cnns_and_gans Goto Github PK

Under which open-source license are you releasing this software and the trained weights ?
I cannot find a License file within the repository.
Thank you for clearing this up

what is the dataset directory?

hey i'm trying to load image 128x128 and i have resized it into 64x64 in data_loader but i still get this error

Traceback (most recent call last):
File "train_gan.py", line 245, in
args.restore_file,cuda_id=cuda_id)
File "train_gan.py", line 127, in train_and_evaluate
train(netG, netD, optimG, optimD, loss_fn, train_dataloader, metrics, params, cuda_id)
File "train_gan.py", line 66, in train
g_loss = loss_fn(fake_out, fake_img, real_img)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\module.py", line 477, in call
result = self.forward(*input, **kwargs)
File "F:\ProgramTA\CGAN_AinilMardiah\program\loss.py", line 78, in forward
perception_loss = self.mse_loss(self.loss_network(out_images), self.loss_network(target_images))
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\module.py", line 477, in call
result = self.forward(*input, **kwargs)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\loss.py", line 421, in forward
return F.mse_loss(input, target, reduction=self.reduction)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\functional.py", line 1716, in mse_loss
return _pointwise_loss(lambda a, b: (a - b) ** 2, torch._C._nn.mse_loss, input, target, reduction)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\functional.py", line 1674, in _pointwise_loss
return lambd_optimized(input, target, reduction)
RuntimeError: input and target shapes do not match: input [16 x 512 x 8 x 8], target [16 x 512 x 2 x 2] at c:\programdata\miniconda3\conda-bld\pytorch_1532505617613\work\aten\src\thcunn\generic/MSECriterion.cu:12

could help me how to solve the error?

Both SRGAN and CGAN's size of best_d.pth.tar are 0. Maybe the file was broken?

The link cannot be opened

I think if loss_D is 0, it is that discriminator can easy to make a distinction between real image and fake image, and this is not good, so why SRGAN loss_D is 0?

Traceback (most recent call last):
File "train_gan.py", line 271, in
train_and_evaluate(netG, netD, train_dl, val_dl, optimG, optimD, loss_fn, metrics, params, opt.model_dir, opt.restore_file,cuda_id=cuda_id)
File "train_gan.py", line 142, in train_and_evaluate
train(netG, netD, optimG, optimD, loss_fn, train_dataloader, metrics, params, cuda_id)
File "train_gan.py", line 75, in train
fake_out = netD(fake_img).mean()
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\module.py", line 477, in call
result = self.forward(*input, **kwargs)
File "F:\ProgramTA\CGAN_AinilMardiah\program\model\cgan.py", line 100, in forward
return torch.sigmoid(self.net(x).view(batch_size))
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\module.py", line 477, in call
result = self.forward(*input, **kwargs)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\container.py", line 91, in forward
input = module(input)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\module.py", line 477, in call
result = self.forward(*input, **kwargs)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\modules\activation.py", line 447, in forward
return F.leaky_relu(input, self.negative_slope, self.inplace)
File "C:\Users\PC\Anaconda3\envs\cgan\lib\site-packages\torch\nn\functional.py", line 755, in leaky_relu
return torch._C._nn.leaky_relu(input, negative_slope)
RuntimeError: CUDA error: out of memory

I have tried to reduce batch size but i still get this error, could you help me how to resolve this error?

in train_gan.py file
in line 15 there should be :
from evaluate_gan import evaluate
instead of "from evaluate import evaluate"
Otherwise, there will be an import error

I want to save the result image. .So, what is the data structure of output_batch ? I don't know how to writte the code ? can you writte it ?

I found tv loss formula in wiki:

while your code shows you did not have a sqrt operation.
could you tell me why?