Traffic flow video analysis method and system based on computer visual identification

By combining the deep learning model YOLOv8 and the optical character recognition technology PaddleOCR, the problem of low recognition accuracy of existing traffic flow monitoring systems in complex environments is solved, real-time and accurate traffic flow analysis is achieved, and system construction and maintenance costs are reduced.

CN119693847APending Publication Date: 2025-03-25BEIJING MINSHENG THINK TANK TECHNOLOGY INFORMATION CONSULTING CO LTD
6 Cites 2 Cited by

Patent Information

Application Number
CN202411806735.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-10
Publication Date
2025-03-25

AI Technical Summary

Technical Problem

The existing traffic flow monitoring systems have problems such as low recognition accuracy, high false alarm rate, high cost, difficult maintenance, and poor identification results in complex backgrounds, nighttime or inclement weather conditions.

Method used

The deep learning model YOLOv8 is combined with the optical character recognition technology PaddleOCR, and the YOLOv8 model is trained through the vehicle image data set, the vehicles in the video are identified and different vehicle types are distinguished. Combined with line collision detection and area detection methods, the vehicle movement trajectory is tracked in real time and traffic flow reports are generated.

Benefits of technology

Maintain high recognition accuracy in complex environments, realize real-time traffic flow analysis, reduce construction costs and maintenance difficulties, and is suitable for various scenarios such as urban roads, highways and parking lots.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to a traffic flow video analysis method and system based on computer visual identification, and the method comprises the steps: 1, employing a YOLOv8 model as a backbone network, training the YOLOv8 model through a marked vehicle image data set, recognizing vehicles in a video through the trained YOLOv8 model, distinguishing different vehicle types, and obtaining a vehicle recognition result; 2, positioning the license plate position of the recognized vehicle through a YOLOv8 model, obtaining a license plate image, and sending the license plate image to PaddleOCR for character recognition to obtain license plate number information; 3, tracking the moving track of the vehicle in real time by using a YOLOv8 model in combination with a line collision detection and region detection method, setting a virtual line or region, and obtaining traffic flow direction information when the vehicle passes through the virtual line or enters a specific region; and step 4, generating a traffic flow report according to the obtained vehicle type information, license plate number information and traffic flow direction information. According to the invention, high identification precision can be maintained in various complex environments.
Need to check novelty before this filing date? Find Prior Art