Add test_image_processing_common.py

[amyeroberts]

What does this PR do?

Creates an equivalent test_feature_extraction_common.py for image processors: test_image_processing_common.py and moves any vision and image processing specific logic to this file.

This is necessary for creating ImageProcessingSavingTestMixin in order to rename any feature extractor references in the image processor tests. This is left for a future PR.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[sgugger]

Shouldn't all the tests use ImageProcessingSavingTestMixin instead of FeatureExtractionSavingTestMixin?

LGTM otherwise, thanks!

[ydshieh]

LGTM, with same question as Sylvain mentioned. Thank you for working on 🖼️ !

You mentioned

This is necessary for creating ImageProcessingSavingTestMixin in order to rename any feature extractor references in the image processor tests. This is left for a #20768.

But the new added image_processing_common.py contains ImageProcessingSavingTestMixin. So it looks a bit confusing, but I believe you have some reasons.

[amyeroberts]

@ydshieh @sgugger Yes, sorry, I could have been clearer above.

The reason for not replacing FeatureExtractionSavingTestMixin with ImageProcessingSavingTestMixin in the test_image_processing_xxx.py files in this PR is that the tests in the original mixin use class attributes like self.feature_extraction_class. When updating the mixin, then attributes in the testing class XxxImageProcessingTest for each test file test_image_processing_xxx.py have to be updated. This results in either 1) updating just the FE references to have the tests running or 2) updating all of the FE references. In the case of 1) it results in the code being mixed between feature extractors and image processors which I found confusing to read (subjective opinion) and for 2) it introduced hundreds of lines of additional diff. I decided to leave the switch to a follow up PR.

[ydshieh]

@ydshieh @sgugger Yes, sorry, I could have been clearer above.

The reason for not replacing FeatureExtractionSavingTestMixin with ImageProcessingSavingTestMixin in the test_image_processing_xxx.py files in this PR is that the tests in the original mixin use class attributes like self.feature_extraction_class. When updating the mixin, then attributes in the testing class XxxImageProcessingTest for each test file test_image_processing_xxx.py have to be updated. This results in either 1) updating just the FE references to have the tests running or 2) updating all of the FE references. In the case of 1) it results in the code being mixed between feature extractors and image processors which I found confusing to read (subjective opinion) and for 2) it introduced hundreds of lines of additional diff. I decided to leave the switch to a follow up PR.

Understand! Thank you for explaining!
BTW, it would be super nice to give a link like

transformers/tests/models/beit/test_image_processing_beit.py

Line 110 in 997cbeb

feature_extraction_class = BeitFeatureExtractor if is_vision_available() else None

(I agree it is a bit more time consuming :-) )