Fix draw_bounding_boxes for tensors on GPU

[sumanthratna]

Background

without this PR, this happens when image is on a CUDA device:

  File ".../my_code.py", line 186, in build_graph
    wsi_with_patches = draw_bounding_boxes(wsi, boxes)  # `wsi` is on CUDA device 0
  File ".../venv/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 28, in decorate_context
    return func(*args, **kwargs)
  File ".../venv/lib/python3.8/site-packages/torchvision/utils.py", line 188, in draw_bounding_boxes
    ndarr = image.permute(1, 2, 0).numpy()
TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

Notes

• not detaching the tensor because the function is decorated with torch.no_grad

[facebook-github-bot]

💊 CI failures summary and remediations

As of commit 858d63a (more details on the Dr. CI page):

---

• 1/1 failures introduced in this PR

---

🕵️ 1 new failure recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

unittest_prototype (1/1)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

�[31m============================== �[31m�[1m6 ...0.31s�[0m�[31m ===============================�[0m

E   ImportError: cannot import name 'ShardingFilter' from 'torch.utils.data.datapipes.iter' (/home/circleci/.local/lib/python3.7/site-packages/torch/utils/data/datapipes/iter/__init__.py)�[0m
------ generated xml file: /home/circleci/project/test-results/junit.xml -------
=========================== short test summary info ============================
ERROR test/test_prototype_builtin_datasets.py
ERROR test/test_prototype_datasets_api.py
ERROR test/test_prototype_datasets_utils.py
ERROR test/test_prototype_features.py
ERROR test/test_prototype_models.py
ERROR test/test_prototype_transforms.py
!!!!!!!!!!!!!!!!!!! Interrupted: 6 errors during collection !!!!!!!!!!!!!!!!!!!!
�[31m============================== �[31m�[1m6 errors�[0m�[31m in 0.31s�[0m�[31m ===============================�[0m


Exited with code exit status 2

---

This comment was automatically generated by Dr. CI (expand for details).

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

[prabhat00155]

Thanks @sumanthratna for the fix. I guess this may have only been tested for CPU tensors only. I see a potential for a similar issue in draw_keypoints(). Do you want to fix that as well?

[sumanthratna]

Thanks @sumanthratna for the fix. I guess this may have only been tested for CPU tensors only. I see a potential for a similar issue in draw_keypoints(). Do you want to fix that as well?

sure, I can fix that issue—should I do it in a separate PR or add it to this one?

[prabhat00155]

sure, I can fix that issue—should I do it in a separate PR or add it to this one?

Feel free to create a separate PR, thanks!

[github-actions]

Hey @prabhat00155!

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

[oke-aditya]

Sorry, but I guess the label for this should be bug fix and not other.