VisorAI

Driver inattention to road markings contributes to a measurable share of urban traffic incidents in the Philippines. Existing ADAS systems are tuned for foreign markings and fail on LTO-specific signage.

1. Curated and hand-labeled a dataset of LTO road markings across day, dusk, and rain conditions.
2. Fine-tuned YOLOv11n on the dataset; quantized for on-device inference.
3. Built a React-Native frontend with a TTS auditory layer that pronounces detected markings without distracting the driver.
4. Throttled inference to 5 FPS on mid-tier devices to stay battery-neutral.