Open-source maintainer building accessible, on-device technology.

I am a graduate student at Xidian University's Guangzhou Research Institute, specializing in Next-Generation Electronic Information Technology, and the primary maintainer of BlindAssist. My work focuses on Android accessibility, on-device machine learning, reproducible evaluation, and evidence-bounded assistive-vision research.

An open-source Android prototype for assistive perception using Kotlin, Jetpack Compose, CameraX, and TensorFlow Lite.

The project makes its Android implementation, build and test entry points, evaluation protocols, research records, and failure boundaries public. It prioritizes local inference, traceable provenance, reproducibility, and honest reporting of negative or inconclusive results.

My maintainer responsibilities include Android and ML engineering, issue triage, release management, contributor guidance, security documentation, and evidence governance.

Repository · Latest release · Contributing · Open-source public value · Security

BlindAssist is an assistive research prototype, not a certified safety device. Its outputs do not replace a white cane, guide dog, professional mobility training, or human judgment.

• Make projects understandable, buildable, testable, and reviewable.
• Keep licenses, model and dataset provenance, and security boundaries explicit.
• Treat unknown evidence as unknown; do not turn experiments into deployment claims.
• Preserve negative results so others can reproduce decisions, not only demos.
• Welcome issues and pull requests with clear contribution paths.

I am interested in collaboration around Android accessibility, assistive technology, on-device computer vision, reproducible evaluation, and open-source maintenance.

For BlindAssist, start with the contribution guide or open an issue.