Skip to main content May 2024 - now May 2024 - now obscuRRRa Offline desktop app for privacy-focused batch face blurring AI/ML ›
- Built Python face blurring app using MTCNN model achieving 96.4% accuracy on 10,000-image dataset
- Enabled batch processing for large-scale face detection and anonymization in local environment
- Developed an offline-capable (GUI) graphic user interface with Tkinter to ensure privacy and usability and released on GitHub
May 2025 - Aug 2025 May 2025 - Aug 2025 Trekeye Rear Hazard Detection Prototype AI/ML ›
- Engineered helmet-mounted hazard detection system with YOLOv8, BoT-SORT/ByteTrack and TTC-based alerts
- Established cross-platform baselines: 18 FPS desktop, 1.8 FPS Raspberry Pi under 2GB RAM; documented negative results
- Delivered reproducible Python framework with YAML configs, CSV/MP4 logging and optical flow for unclassified hazards