AUTOMATED ROAD ASSESSMENT SYSTEM

Developed a real-time road defect detection system capable of identifying potholes and line cracks using convolutional neural networks (CNNs) and advanced image processing techniques with TensorFlow and OpenCV. The system analyzes live video feeds from vehicle-mounted cameras and achieved 93% pothole and 89% crack detection accuracy under varying lighting and weather conditions. It includes GPS-based location tagging, defect size estimation, and a centralized report management system with status tracking through an admin dashboard. The platform was deployed on Render with a custom domain, and features heatmap-based visualization to identify high-priority defect zones, supporting efficient maintenance planning and government reporting.