fix torch backend resize with pad_to_aspectio_ratio is set to True

[sineeli]

The resize doesn't work as expected while setting flag pad_to_aspectio_ratio=True.

1. Resize with JAX works as expected.
2. Resize with Torch needs fix.
3. Resize with Torch Fixed Code

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 67.85714% with 9 lines in your changes missing coverage. Please review.

Project coverage is 82.01%. Comparing base (5df8fb9) to head (41ecb92).
Report is 6 commits behind head on master.

Files with missing lines	Patch %	Lines
keras/src/backend/numpy/image.py	66.66%	3 Missing and 3 partials ⚠️
keras/src/backend/torch/image.py	70.00%	1 Missing and 2 partials ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #20797    +/-   ##
========================================
  Coverage   82.01%   82.01%            
========================================
  Files         557      558     +1     
  Lines       52014    52150   +136     
  Branches     8037     8062    +25     
========================================
+ Hits        42657    42771   +114     
- Misses       7403     7416    +13     
- Partials     1954     1963     +9     

Flag	Coverage Δ
keras	81.83% <67.85%> (+<0.01%)	⬆️
keras-jax	64.26% <0.00%> (+0.04%)	⬆️
keras-numpy	58.99% <42.85%> (+0.04%)	⬆️
keras-openvino	29.86% <0.00%> (-0.04%)	⬇️
keras-tensorflow	64.81% <0.00%> (+0.02%)	⬆️
keras-torch	64.19% <25.00%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fchollet]

Thanks for the PR -- can you add a simple unit test for this change?

[sineeli]

Thanks for the PR -- can you add a simple unit test for this change?

Added a tests under image_test.py with pad_to_aspect_ratio=True

[fchollet]

Thanks a lot! Can you take a look at the resulting test failure with the numpy backend? Looks like a numerical mismatch, perhaps numpy had the same issue. https://github.com/keras-team/keras/actions/runs/12956346950/job/36142411678?pr=20797

[sineeli]

Thanks a lot! Can you take a look at the resulting test failure with the numpy backend? Looks like a numerical mismatch, perhaps numpy had the same issue. https://github.com/keras-team/keras/actions/runs/12956346950/job/36142411678?pr=20797

Yup it seems like that, logic of numpy backend is also off just corrected it.

[fchollet]

Nice, thank you for the fix!