Open morrisalp opened 7 months ago
Hi @morrisalp,
I checked from my end, the file is there after I downloaded it. Is your folder hierarchy the same as shown in the README.md
?
Also, it looks like a similar problem as your other issue, so try adding the project to PYTHONPATH
(see README.md
).
Regardless, you don't have to use this image, you can instead create a checkerboard directly in the code -- so you can replace this line https://github.com/dolev104/neural_congealing/blob/main/utils_atlas/logger.py#L256 (also appears here) with the following code:
import torch
from torchvision import transforms
squares_per_side = 32
checkerboard_lowres = torch.eye(2).unsqueeze(0).repeat(3, squares_per_side // 2, squares_per_side // 2) * 0.2 + 0.6
checkerboard = transforms.Resize((256, 256), interpolation=transforms.InterpolationMode.NEAREST)(checkerboard_lowres)
checkerboard = checkerboard[None, ...].to(self.device)
Let me know if you're still having problems with anything.
I see, I didn't realize downloading the sample image sets (containing this file) was required for this. Thanks!
Running train.py throws the error:
FileNotFoundError: [Errno 2] No such file or directory: 'data/checkerboard_gray_256.png'
. It seems like this missing file is required bypropagate_edit.py
andlogger.py
.