Accident Detection and Alert System Using Deep Learning

Road accident is a major concern around the world, causing loss of life and important goods. The need to detect accidents in real time and emergency intervention has become more important than ever. This project isdeveloping an Accident Detection and Alert System using deep learning techniques to improve road safety and reduce response time. Yolov8, an object detection model is used to detect accidents in real time and determine the license plate of the car involved. After the accident is being discovered, PaddleOCR extracts and recognizes the characters and numbers on the plate. The number of licenses is then combined with the database to restore the details of the vehicle owner, allowing automatic emergency alerts on the guardian and the nearest medical response group. The proposed system eliminates the need of manual intervention, ensuring faster support. By utilizing CCTV footage and integrating deep learning techniques, it provides a saclable, cost-effective and automated approach to accident detection and reporting. Experimental results demonstarte that YOLOv8 achieves high accuracy in detecting accidents and extracting license plates, while PaddleOCR esures reliable character recognition across various license plate formats and lighting conditions.