Hao
If your images are 3D, the APIs would not return 2D images ... If it happen, I can't help ...
This one is necessary, the new API uses `model.train()` `model.eval()` for train/test switching.
make sure your input is dtype=np.float32
``` tensorflow.python.framework.errors_impl.InvalidArgumentError: cannot compute Conv2D as input #1(zero-based) was expected to be a double tensor but is a float tensor [Op:Conv2D] name: conv2d_1 ``` this error only happen when the...
I add a line here to force all images to be float32, hope it helps https://github.com/tensorlayer/srgan/blob/master/train.py#L332
thanks for the issue, apart from the warning, is there any thing else need to be improved?
@AndrzejBandurski set `persistent=True`
download it.
it is too small, we can't crop 384x384 subimages from 256x256 images. You should use image with size around 1000 to 2000
yes, you can have a try