Resize 96*96*3 image in STL10 dataset into 48*48*3 image

1

I use the following codes, but it doesn’t work.

dataset0 = torchvision.datasets.STL10('./Drive/training/stl10', split='train+unlabeled', download=True,
                                     transform=torchvision.transforms.Compose([
                                         torchvision.transforms.ToTensor(),
                                            torchvision.transforms.Resize(64),
                                     ]))

also tried
torchvision.transforms.Resize(int(64)),

I find that there is another parameter,
Is there anyone can give me some suggestions?

torchvision.transforms. Resize (size, interpolation=2 )

interpolation (int, optional) – Desired interpolation enum defined by filters. Default is PIL.Image.BILINEAR. If input is Tensor, only PIL.Image.NEAREST, PIL.Image.BILINEAR and PIL.Image.BICUBIC are supported.

2

You need to put the resize before the ToTensor() transform.

1 Like
3

Thank you for your declearation