Convolutional neural network-based retrograde vehicle detection method and system, and medium

Abstract

The invention discloses a convolutional neural network-based retrograde vehicle detection method. The method comprises the following steps of obtaining traffic monitoring video data; performing real-time target detection on the traffic monitoring video by adopting the trained YOLOv3 neural network model to obtain a detection target; performing target tracking on the detection target by adopting anSORT method, and recording an ID and a detection frame of each target vehicle; calculating the positions of different detection frames generated under different frame numbers of each target vehicle to obtain the driving direction of the vehicle, and judging whether the vehicle runs reversely or not; if so, capturing a picture; or otherwise, returning to the step of continuously performing targettracking on the detected vehicle by adopting the SORT method. According to the detection method, real-time multi-target tracking is realized, the driving direction of the vehicle is automatically identified, whether the vehicle violates rules and regulations is accurately judged, and snapshot evidence obtaining is performed on the violating vehicle.

Description

technical field [0001] The present invention relates to the technical field of computer vision, in particular to a convolutional neural network-based detection method, system, terminal and medium for retrograde vehicles. Background technique [0002] Vehicle detection and tracking is one of the important directions of artificial intelligence research. It has many applications in real life. The detection and tracking work is completely manual, which requires a lot of manpower, material and financial resources. Using the vehicle detection and tracking technology based on deep learning can quickly and accurately detect and track illegal vehicles. [0003] With the rapid development of economy and urbanization, the total number of roads and vehicles in various cities in our country continues to increase, and the management pressure of traffic control departments is increasing day by day. Although high-definition surveillance cameras have been deployed at most intersections, the...

AI Technical Summary

Problems solved by technology

Although high-definition surveillance cameras have been deployed at most intersections, the amount of video generated every day is also increasing. It is time-consuming and laborious to detect and chase videos in real time manually, and is prone to delays and omissions. Therefore, it is urgent to find a An automated method assists manual monitoring and processing, which is the core of current intelligent transportation

Images