New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Different settings in `random_sized_crop` #134

Open

wjfwzzc opened this issue Apr 19, 2021 · 0 comments

wjfwzzc commented Apr 19, 2021

In pycls, aspect_ratio are sampled uniformly in linear space:

pycls/pycls/datasets/transforms.py

Line 38 in 2c361ee

aspect_ratio = np.random.uniform(3.0 / 4.0, 4.0 / 3.0)

However in torchvision, aspect_ratio are sampled uniformly in log space:
https://github.com/pytorch/vision/blob/79ee22da44d43dabcc0fd30da2c914c5f04143eb/torchvision/transforms/transforms.py#L849-L854

So does timm:
https://github.com/rwightman/pytorch-image-models/blob/e8a64fb88108b592da192e98054095b1ee25e96e/timm/data/transforms.py#L106-L107

Seems that issue #123 reports the performance decrease. Could anyone check that?

The text was updated successfully, but these errors were encountered:

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment