Skip to content

added new overlay classes for enhanced image robustness#1141

Merged
chakravarthik27 merged 5 commits intorelease/2.5.0from
feature/random-masking-on-images-tests
Nov 18, 2024
Merged

added new overlay classes for enhanced image robustness#1141
chakravarthik27 merged 5 commits intorelease/2.5.0from
feature/random-masking-on-images-tests

Conversation

@chakravarthik27
Copy link
Collaborator

This pull request introduces several enhancements and new features to the langtest/transform/image/robustness.py file. The changes focus on improving image transformation robustness and adding new classes for various image overlay techniques.

Enhancements to existing transformations:

  • Image Text Overlay: Updated default text, font size, and position for better visibility and added transparency to the font color.
  • Image Watermark: Added a default value for the watermark parameter to simplify usage.

New features:

  • ImageRandomTextOverlay class: Introduced a new class to add random text overlays on images, enhancing robustness testing for visual QA tasks.
  • ImageRandomLineOverlay class: Added a new class to overlay random lines on images, providing another method for robustness testing in visual QA.
  • ImageRandomPolygonOverlay class: Created a new class to overlay random polygons on images, further diversifying robustness testing methods for visual QA.

Minor improvements:

  • Reset mask in transform method: Ensured the mask is reset after each iteration to maintain consistency in transformations. [1] [2]

@chakravarthik27 chakravarthik27 self-assigned this Nov 18, 2024
@chakravarthik27 chakravarthik27 merged commit 12cc07e into release/2.5.0 Nov 18, 2024
@chakravarthik27 chakravarthik27 linked an issue Dec 24, 2024 that may be closed by this pull request
@chakravarthik27 chakravarthik27 deleted the feature/random-masking-on-images-tests branch December 24, 2024 16:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Random Masking on Images Tests

2 participants