Enable to set random ranges to RandomEmojiOverlay parameters

[lyakaap]

Summary

• I have read CONTRIBUTING.md to understand how to contribute to this repository :)

Added randomness to RandomEmojiOverlay parameters like emoji_size.
A motivation is that I wanted to randomly change the parameters on-the-fly, rather than "fixed" like current implementation.
This modified augmentation was actually used in ISC21 Descriptor Track 1st-place solution.

Test Results

The tests are passed except the following error.

Traceback (most recent call last):
  File "/Users/shuhei.yokoo/Documents/AugLy/augly/tests/image_tests/transforms_unit_test.py", line 177, in test_RandomEmojiOverlay
    self.evaluate_class(
  File "/Users/shuhei.yokoo/Documents/AugLy/augly/tests/image_tests/base_unit_test.py", line 129, in evaluate_class
    are_equal_images(dst, ref), "Expected and outputted images do not match"
  File "/Users/shuhei.yokoo/Documents/AugLy/augly/tests/image_tests/base_unit_test.py", line 20, in are_equal_images
    return a.size == b.size and np.allclose(np.array(a), np.array(b))
  File "<__array_function__ internals>", line 5, in allclose
  File "/Users/shuhei.yokoo/.pyenv/versions/augly/lib/python3.8/site-packages/numpy/core/numeric.py", line 2249, in allclose
    res = all(isclose(a, b, rtol=rtol, atol=atol, equal_nan=equal_nan))
  File "<__array_function__ internals>", line 5, in isclose
  File "/Users/shuhei.yokoo/.pyenv/versions/augly/lib/python3.8/site-packages/numpy/core/numeric.py", line 2358, in isclose
    return within_tol(x, y, atol, rtol)
  File "/Users/shuhei.yokoo/.pyenv/versions/augly/lib/python3.8/site-packages/numpy/core/numeric.py", line 2339, in within_tol
    return less_equal(abs(x-y), atol + rtol * abs(y))
ValueError: operands could not be broadcast together with shapes (1080,1920,3) (1080,1920,4) 

----------------------------------------------------------------------
Ran 71 tests in 35.609s

FAILED (errors=1, skipped=4)
sys:1: ResourceWarning: unclosed file <_io.BufferedReader name='/Users/shuhei.yokoo/Documents/AugLy/augly/assets/tests/image/inputs/dfdc_1.jpg'>

It seems that expected image has alpha channel whereas output image doesn't have alpha channel.
I'm not sure why it happened. Replacing the expected image is needed?

[zpapakipos]

Hi @lyakaap, thanks so much for this PR! :)

I left a couple comments with small changes I'd like before landing this. You will indeed need to replace the expected output file for the RandomEmojiOverlay unit test. You can do this e.g. by temporarily adding a line pil_dst.save("augly/assets/tests/image/dfdc_expected_output/test_RandomEmojiOverlay.png") here and then running python -m unittest augly.tests.image_tests.functional_unit_test.FunctionalImageUnitTest.test_RandomEmojiOverlay. Then if that passes, remove the line you added in base_unit_test.py and commit your changes (should just be the new test_RandomEmojiOverlay.png file)!

[lyakaap]

Thank you for your quick & kind review!

I reflected your review and confirmed the tests are passed.
Like you instructed, I rewrote png file by inserting the line, but in evaluate_class instead evaluate_function.
And executed python -m unittest augly.tests.image_tests.transforms_unit_test.TransformsImageUnitTest.test_RandomEmojiOverlay.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[zpapakipos]

Sorry for the delay with reviewing this PR, I was on PTO for the holidays! :) Please fix the nits I suggested below, and also run black on the files which are failing in the lint workflow, as you can see.

black augly/image/functional.py
black augly/tests/image_tests/transforms_unit_test.py

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[facebook-github-bot]

@lyakaap has updated the pull request. You must reimport the pull request before landing.

[facebook-github-bot]

@zpapakipos has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.