AI_Traffic_COP
AI Traffic Cop is a real-time traffic violation detection system using computer vision to identify helmet and seatbelt violations, track vehicles, and recognize license plates from video feeds.
Videos
Description
AI Traffic Cop is a real-time traffic monitoring system that currently detects bikes and identifies helmet violations from video feeds. Additional features like seatbelt and license plate detection will be added soon.
Progress During Hackathon
During the hackathon, our team focused on laying the foundation for "AI Traffic Cop" — a real-time traffic violation detection system powered by machine learning and computer vision. We successfully completed the core functionality of detecting bikes and identifying helmet violations using a YOLOv8 object detection model. This involved collecting and preprocessing helmet detection datasets, training and fine-tuning the model, and integrating it with a video processing pipeline using OpenCV. A dedicated script was built to process uploaded videos frame-by-frame, detect riders and helmets, and flag violations where helmets were missing. We also implemented a basic logging system that records each violation with a timestamp and saves cropped frames of the violators. Although our initial goal included detecting seatbelt violations, red light jumping, and license plates, due to time constraints, we prioritized helmet violation detection — which now works reliably and in real time. We're planning to expand the system post-hackathon by integrating: Seatbelt detection using a custom CNN classifier License plate recognition using EasyOCR Vehicle tracking with DeepSORT or ByteTrack to handle multiple violations simultaneously Overall, we made solid progress on a key safety feature and built a scalable base to continue development beyond the hackathon.