Profile Information
Name
Tangil Hossain Shawon
IRC nickname on libera.chat IRC
(Not set yet; I am willing to join and participate actively if required)
GitHub: https://github.com/TangilHossain
Resume
Location Bangladesh
Typical working hours (include your timezone)
I typically work between 10:00 AM – 6:00 PM (UTC+6). I can adjust my schedule when needed to align with mentor availability.
Synopsis
This project aims to implement lossless JPG transformations in the Wikimedia Commons Android app, enabling users to edit images directly within the app without degrading quality or losing metadata.
A key issue in the current pipeline is that while pixel-level transformations (via LLJTran) are lossless, metadata handling is not always preserved correctly. For example, using Android’s ExifInterface.saveAttributes() after rotation can rewrite the metadata block and strip out important segments like the ICC color profile, leading to visible color or brightness shifts.
This project will address such issues and extend the editing capabilities by:
- Ensuring fully lossless rotation and cropping, preserving both image data and metadata
- Implementing lossless blur by operating at the JPEG 8×8 block level, avoiding full recompression
- Adding optional automatic face and license plate blurring for privacy protection
The expected outcome is a seamless in-app editing experience where users can safely modify images without relying on third-party tools, improving usability and increasing contribution rates.
Possible Mentor(s)
@Nicolas_Raoul
@RitikaPahwa4444
Have you contacted your mentors already?
I have initiated communication through GitHub discussions and relevant issues to get early feedback on my approach.
Deliverables
Community Bonding Period
- Set up and explore the Commons Android app codebase
- Study JPEG internals (DCT blocks, EXIF, ICC profiles, APP segments)
- Review LLJTran usage and current transformation pipeline
- Discuss approach and expectations with mentors
Week 1–2: Investigation & Planning
- Reproduce known issues such as ICC profile loss after transformations
- Analyze how metadata is currently handled
- Validate behavior using tools like exiftool
- Finalize implementation approach for lossless transformations and blur
Milestone: Clear and validated technical plan
Week 3–5: Fix Lossless Rotation & Metadata Handling
- Remove or replace APIs that rewrite metadata destructively (e.g., ExifInterface where inappropriate)
- Ensure ICC profiles and EXIF metadata are preserved correctly
- Validate using byte-level comparison and visual checks
- Add unit tests for metadata integrity
Milestone: Fully reliable lossless rotation/cropping pipeline
Week 6–8: Implement Lossless Blur
- Implement region-based blur using JPEG 8×8 block structure
- Ensure only selected blocks are modified
- Avoid recompression of unaffected regions
- Provide manual selection support
Milestone: Functional selective blur without global recompression
Week 9–10: Automatic Detection for Privacy
- Integrate face detection (e.g., ML Kit or similar)
- Detect license plates if feasible
- Automatically apply blur to detected regions
- Allow user adjustments
Milestone: Working auto-blur feature
Week 11–12: UI Integration & Optimization
- Integrate editing features into the app UI
- Add preview and usability improvements
- Optimize performance for mobile devices
Milestone: Smooth and user-friendly editing experience
Week 13: Testing & Documentation
- Add unit and integration tests
- Perform real-device validation
- Document implementation and usage
Milestone: Final submission ready
Participation
I plan to actively communicate and collaborate throughout the program:
- Share regular updates through GitHub issues and pull requests
- Ask questions and seek feedback early when needed
- Follow Wikimedia’s contribution and code review practices
- Be responsive to mentor feedback and iterate accordingly
All development work will be done transparently in the Commons Android repository.
About Me
I am currently a student of CSE at CUET and actively learning Android development with Kotlin, focusing on building a strong foundation by working on real-world problems and exploring open-source projects.
I learned about GSoC through developer communities and was particularly interested in Wikimedia due to its global impact and open knowledge mission.
I will be able to dedicate consistent time to this project during the program and do not have any major conflicting commitments.
I am primarily applying to GSoC with Wikimedia for this cycle.
This project is especially meaningful to me because it combines low-level image processing concepts with practical mobile development, and solving it would directly improve user experience and privacy for contributors worldwide.
Past Experience
While I have not yet contributed to the Wikimedia Commons app, I have started exploring the codebase and analyzing relevant issues.
For example, I examined the image rotation pipeline and identified that using ExifInterface.saveAttributes() after LLJTran transformations may lead to metadata loss, particularly the ICC color profile, which explains the observed brightness shifts. This helped me understand how metadata handling can affect visual output even when pixel data remains unchanged.
I have experience with:
Kotlin and Android development
Firebase integration in mobile apps
Debugging and understanding existing codebases
Through my work, I have developed the ability to:
Break down complex technical problems
Learn new concepts (such as JPEG internals) as needed
Write clean and maintainable code
GitHub: https://github.com/TangilHossain
Any Other Info
Relevant issues:
https://github.com/commons-app/apps-android-commons/issues/6659
https://github.com/commons-app/apps-android-commons/issues/190
https://github.com/commons-app/apps-android-commons/issues/1192
Planned approach:
- Use tools like exiftool for validation
- Ensure byte-level preservation of metadata where required
- Prototype features (like blur) before full integration
I plan to start contributing small fixes and improvements to the codebase during the application period to better understand the system and workflow.