Preprocessing used for ImageNet-pretrained models

What transforms (random crops, flips, etc.) were applied to the training data for the standard imagenet-pretrained models (vgg-16, alexnet, etc.) available through the model zoo?

1 Like

Did you ever find the answer to this?

I believe this:

train_loader = torch.utils.data.DataLoader(
        datasets.ImageFolder(traindir, transforms.Compose([
            transforms.RandomSizedCrop(224),
            transforms.RandomHorizontalFlip(),
            transforms.ToTensor(),
            normalize,
        ])),
        batch_size=args.batch_size, shuffle=True,
        num_workers=args.workers, pin_memory=True)

    val_loader = torch.utils.data.DataLoader(
        datasets.ImageFolder(valdir, transforms.Compose([
            transforms.Scale(256),
            transforms.CenterCrop(224),
            transforms.ToTensor(),
            normalize,
        ])),
        batch_size=args.batch_size, shuffle=False,
        num_workers=args.workers, pin_memory=True)

From https://github.com/pytorch/examples/blob/42e5b996718797e45c46a25c55b031e6768f8440/imagenet/main.py#L95