allow size to be generic Sequence

[wangg12]

For example, when we pass a size arg in a config created by omegaconf, only allowing list, tuple will raise an error.

cc @vfdev-5

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/7999

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 3 New Failures, 11 Unrelated Failures

As of commit 0fd300c with merge base a8c4875 ():

NEW FAILURES - The following jobs have failed:

• Build M1 Wheels (gh)
• Build Macos Wheels (gh)
• unittests-macos (3.8, macos-m1-12) / macos-job (gh)

FLAKY - The following jobs failed but were likely due to flakiness present on trunk:

• macos (macos-12) / macos-job (gh)
• macos (macos-m1-12) / macos-job (gh)
• unittests-macos (3.10, macos-12) / macos-job (gh)
• unittests-macos (3.11, macos-12) / macos-job (gh)
• unittests-macos (3.8, macos-12) / macos-job (gh)
• unittests-macos (3.9, macos-12) / macos-job (gh)
• unittests-windows (3.10, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.11, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.8, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.9, windows.4xlarge, cpu) / windows-job (gh)
• windows (windows.4xlarge, cpu) / windows-job (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[NicolasHug]

Thanks @wangg12 , LGTM (I'll self-commit my suggestion below and wait for the CI before merging)

Technically we should probably be using Sequence from collections.abc instead of from typing (EDIT: because most of the typing stuff is deprecated from 3.9), but maybe this is better addressed in a more general PR. @pmeier any opinion?

[pmeier]

Technically we should probably be using Sequence from collections.abc instead of from typing, but maybe this is better addressed in a more general PR.

typing.Sequence is actually an alias for collections.abc.Sequence. Meaning, we can do it for clarity, but there is no functional difference.

---

Should we also fix the error message below to reflect that this supports arbitrary sequences now?

vision/torchvision/transforms/v2/_geometry.py

Lines 140 to 143 in 1a9ff0d

	else:
	raise ValueError(
	f"size can either be an integer or a list or tuple of one or two integers, " f"but got {size} instead."
	)

[NicolasHug]

Linux unittests and linters are happy, merging. Thanks @wangg12 !