A Vision-Based System: Detecting Traffic Law Violation Case Study of Red-Light Running Using Pre-Trained YOLOv8 Model and OpenCV

Object detection represents a computer vision methodology utilized for recognizing objects within images or videos. Through various algorithms, computers have achieved the capability to identify objects in images. One notable algorithm is YOLOv8, recently unveiled by Ultralytics on January 10, 2023. Ultralytics provides pre-trained models trained on the Microsoft COCO 2017 dataset across multiple scales: nano, small, medium, large, and extra-large. This study presents an investigation into the deployment of the pre-trained YOLOv8 model for the detection of traffic violations, specifically focusing on instances where drivers run red lights. Additionally, this research leverages the OpenCV library to ascertain the color of traffic lights. Consequently, our system detects and quantifies instances of violations. This study lays the groundwork for developing a tool capable of identifying various traffic law violations, documenting daily violation counts, and employing the accumulated data as a dataset for Machine Learning to discern patterns of violations on specific roadways.