add tests for F.pad_bounding_box #6038
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
8 tasks
|
@pmeier a correctness test on cpu/cuda could be good to have as well |
Agreed, but I'm covered right now. Can we add that after #5879? Otherwise, would you be able to contribute it? |
|
@pmeier OK let me merge this PR and add a correctness test |
pmeier
commented
May 26, 2022
Comment on lines
+1140
to
+1147
| bbox = convert_bounding_box_format( | ||
| bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format, copy=False | ||
| ) | ||
| if bbox.dtype != bbox_dtype: | ||
| # Temporary cast to original dtype | ||
| # e.g. float32 -> int | ||
| bbox = bbox.to(bbox_dtype) | ||
| return bbox |
There was a problem hiding this comment.
- There is no need for this
if:.tois a no-op if no conversion needs to be performed. - You don't need to use the dtype explicitly. You can pass a tensor to
.toin which case it will convert device and dtype to the input.
Suggested change
| bbox = convert_bounding_box_format( | |
| bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_format, copy=False | |
| ) | |
| if bbox.dtype != bbox_dtype: | |
| # Temporary cast to original dtype | |
| # e.g. float32 -> int | |
| bbox = bbox.to(bbox_dtype) | |
| return bbox | |
| return convert_bounding_box_format( | |
| bbox, old_format=features.BoundingBoxFormat.XYXY, new_format=bbox_.format, copy=False | |
| ).to(bbox_) |
There was a problem hiding this comment.
@pmeier actually it does not work with .to between tensors and BoundingBox:
>>> t = torch.tensor([[1.0, 2.0, 3.0, 4.0]])
>>> bbox = features.BoundingBox([[0, 1, 2, 3]], format=features.BoundingBoxFormat.XYXY, image_size=(32, 32))
>>> t.to(bbox)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/vision/torchvision/prototype/features/_feature.py", line 83, in __torch_function__
return cls.new_like(args[0], output, dtype=output.dtype, device=output.device)
File "/vision/torchvision/prototype/features/_bounding_box.py", line 54, in new_like
format=format if format is not None else other.format,
AttributeError: 'Tensor' object has no attribute 'format'I'll keep my code, especially it is a temporary cast once we fixed convert_bounding_box_format we can remove it quickly
There was a problem hiding this comment.
I'll added #6094 to track this.
vfdev-5
approved these changes
May 26, 2022
facebook-github-bot
pushed a commit
that referenced
this pull request
Jun 1, 2022
Summary: * add tests for F.pad_bounding_box * Added correctness tests for pad and reimplemented bbox op to keep dtype * Update _geometry.py Reviewed By: NicolasHug Differential Revision: D36760926 fbshipit-source-id: 6c2a5430790e2db778911dbf62ce8e38f6e9b603 Co-authored-by: vfdev <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related to #5514 as it is the last missing item.