RoadLens AI — Intelligent Road Perception

RoadLens AI is an AI-powered road perception and road intelligence platform that transforms road-scene images into structured, machine-readable insights for ADAS and autonomous-driving research. The system implements an end-to-end perception pipeline covering image-quality assessment, adaptive preprocessing, object and traffic-sign detection, modular OCR, confidence estimation, and semantic interpretation. Instead of returning only bounding boxes, RoadLens converts visual observations into meaningful road intelligence such as detected speed limits, stop signs, warning signs, work zones, and license-plate observations. A key feature is its country-aware perception layer, currently supporting USA and India profiles. The USA profile is aligned with the challenge benchmark and focuses on traffic signs, speed-limit signs, advisory speed plaques, work-zone warnings, and license plates. The India profile adds India-specific speed-unit interpretation and rule-based license-plate format validation. RoadLens AI also includes an adverse-condition assessment module that evaluates factors such as sharpness, illumination, sensor noise, glare, and perspective to provide additional context about image reliability. The project combines Python, FastAPI, PyTorch, YOLO, OpenCV, React, and modular OCR components, with model development and benchmarking performed on AMD GPU infrastructure. RoadLens AI is designed as a research and prototyping perception layer. It does not directly control vehicle steering, braking, or acceleration.