-
Notifications
You must be signed in to change notification settings - Fork 7k
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.
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
fix test_detection_preset for ssd data augmentation #7447
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/7447
Note: Links to docs will display an error until the docs builds have been completed. This comment was automatically generated by Dr. CI and updates every 15 minutes. |
@@ -1875,7 +1875,7 @@ def test_detection_preset(image_type, data_augmentation, to_tensor, sanitize): | |||
elif data_augmentation == "ssd": | |||
t = [ | |||
transforms.RandomPhotometricDistort(p=1), | |||
transforms.RandomZoomOut(fill=defaultdict(lambda: (123.0, 117.0, 104.0), {datapoints.Mask: 0})), | |||
transforms.RandomZoomOut(fill=defaultdict(lambda: (123.0, 117.0, 104.0), {datapoints.Mask: 0}), p=1), |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Driveby. All other transforms already use p=1
. Prior to #7442, by chance this transform was not applied.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks Philip
Hey @NicolasHug! You merged this PR, but no labels were added. The list of valid labels is available at https://github.com/pytorch/vision/blob/main/.github/process_commit.py |
Reviewed By: vmoens Differential Revision: D44416577 fbshipit-source-id: 3951020abaf72e0634fc96159bd860c0a17aa8e9
#7442 broke this test. However, this is not a bug in the PR, but rather me forgetting to fix the test setup. #7442 changed the random sampling strategy (see #7442 (comment)), which ultimately leads to a different number of boxes in the final output. As expected, only keeping
RandomPhotometricDistort
in the pipeline yields exactly the same results prior and after #7442.cc @vfdev-5