Skip to content
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

Stricter SanitizeBoundingBoxes labels_getter heuristic #7880

Merged
merged 1 commit into from
Aug 24, 2023
Merged

Stricter SanitizeBoundingBoxes labels_getter heuristic #7880

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 3 additions & 4 deletions gallery/v2_transforms/plot_transforms_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,10 +99,9 @@
format="XYXY", canvas_size=img.shape[-2:])

transforms = v2.Compose([
v2.RandomPhotometricDistort(),
v2.RandomIoUCrop(),
v2.RandomHorizontalFlip(p=0.5),
v2.SanitizeBoundingBoxes(),
v2.RandomResizedCrop(size=(224, 224), antialias=True),
v2.RandomPhotometricDistort(p=1),
v2.RandomHorizontalFlip(p=1),
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was actually mis-using SanitizeBoundingBoxes() as I wasn't passing a label, so I should have set labels_getter to None (see non-regression test below). I decided to just remove SanitizeBoundingBoxes here because that would require making the input a dict, also forcing to change the narrative below. It's OK, SanitizeBoundingBoxes will be clearly used and mentioned in the e2e example that I'm working on now.

])
out_img, out_bboxes = transforms(img, bboxes)

Expand Down
14 changes: 14 additions & 0 deletions test/test_transforms_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1256,6 +1256,20 @@ def test_sanitize_bounding_boxes(min_size, labels_getter, sample_type):
assert out_labels.tolist() == valid_indices


def test_sanitize_bounding_boxes_no_label():
# Non-regression test for https://github.com/pytorch/vision/issues/7878

img = make_image()
boxes = make_bounding_boxes()

with pytest.raises(ValueError, match="or a two-tuple whose second item is a dict"):
transforms.SanitizeBoundingBoxes()(img, boxes)

out_img, out_boxes = transforms.SanitizeBoundingBoxes(labels_getter=None)(img, boxes)
assert isinstance(out_img, datapoints.Image)
assert isinstance(out_boxes, datapoints.BoundingBoxes)


def test_sanitize_bounding_boxes_errors():

good_bbox = datapoints.BoundingBoxes(
Expand Down
2 changes: 1 addition & 1 deletion torchvision/transforms/v2/_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,7 +112,7 @@ def _find_labels_default_heuristic(inputs: Any) -> torch.Tensor:
inputs = inputs[1]

# MixUp, CutMix
if isinstance(inputs, torch.Tensor):
if is_pure_tensor(inputs):
return inputs

if not isinstance(inputs, collections.abc.Mapping):
Expand Down
Loading