transforms.Solarize probability is zero for transforms but not transforms_prime

[dewball345]

I was viewing the transforms class and I saw this

        self.transform = transforms.Compose([
            transforms.RandomResizedCrop(224, interpolation=Image.BICUBIC),
            transforms.RandomHorizontalFlip(p=0.5),
            transforms.RandomApply(
                [transforms.ColorJitter(brightness=0.4, contrast=0.4,
                                        saturation=0.2, hue=0.1)],
                p=0.8
            ),
            transforms.RandomGrayscale(p=0.2),
            GaussianBlur(p=1.0),
            Solarization(p=0.0),
            transforms.ToTensor(),
            transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                 std=[0.229, 0.224, 0.225])
        ])
        self.transform_prime = transforms.Compose([
            transforms.RandomResizedCrop(224, interpolation=Image.BICUBIC),
            transforms.RandomHorizontalFlip(p=0.5),
            transforms.RandomApply(
                [transforms.ColorJitter(brightness=0.4, contrast=0.4,
                                        saturation=0.2, hue=0.1)],
                p=0.8
            ),
            transforms.RandomGrayscale(p=0.2),
            GaussianBlur(p=0.1),
            Solarization(p=0.2),
            transforms.ToTensor(),
            transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                 std=[0.229, 0.224, 0.225])
        ])

Notice that the solarization probability is zero on the first augmentation but not the second. Is there a difference between the two(their different naming?)

Is this expected?

[jzbontar]

We used the same data augmentation as BYOL. They use solarization in only one of the branches.

https://github.com/deepmind/deepmind-research/blob/5f6b11299e93b0d6c6edc8e89cd1dde61495b9ae/byol/utils/augmentations.py#L50

[dewball345]

Are there any benefits to doing so?

…

On Tue, Oct 26, 2021 at 10:59 PM Jure Zbontar ***@***.***> wrote: Closed #35 <#35>. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#35 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHGJKVHZU7QXBO4XYQMO63TUI6PN7ANCNFSM5GEEIJPQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.

[jzbontar]

I didn't run that experiment so I can't really say, sorry.