Video perception-based vehicle abnormal driving detection method

Abstract

The present invention relates to a video perception-based vehicle abnormal driving detection method, comprising: collecting continuous frame sequence image data of a vehicle from road monitoring equipment; adopting a YOLO algorithm to detect the vehicle in the continuous frame sequence image data and obtaining a vehicle detection result; wherein the vehicle detection result comprises: a bounding box position and a category label of each vehicle; screening an abnormal motion candidate set based on the vehicle detection result; confirming a final abnormal behavior event based on the abnormal motion candidate set; wherein the abnormal behavior event comprises: retrograde movement or lane crossing behavior. The present invention significantly improves accuracy of abnormal behavior recognition and provides an efficient and reliable solution for intelligent traffic management.

Images