agaldran / torchvision_paired_transforms Goto Github PK

Hey,

I was excited to find your repo. Sadly I have got the following problem:

do you have any idea why the size of the 2nd image tensor is only 2D?

=============transforms============================

transform_list = [tr.ToTensor(),
tr.Normalize((0.5, 0.5, 0.5),
(0.5, 0.5, 0.5))]
transform_list = [tr.RandomCrop((height, width))] + transform_list
transform_list = [tr.Resize(new_size)] + transform_list
transform_list = [tr.RandomHorizontalFlip()] + transform_list
transform = tr.Compose(transform_list)

img_tr1, img_tr2 = transform(test_img, test_img)
print(img_tr1.size())
print(img_tr2.size())

======output=========================================
torch.Size([3, 512, 512])
torch.Size([512, 512])