Target type identification method and device, server and computer readable storage medium

Abstract

The invention discloses a target type identification method and device, a server and a computer readable storage medium, and the method comprises the steps: obtaining a to-be-identified video, and extracting an image frame in the to-be-identified video according to a preset time interval to serve as a first to-be-identified image; calculating the first to-be-recognized image by using a detection tracking algorithm to obtain a second to-be-recognized image containing each target; inputting the second to-be-recognized image into a target detection model to obtain a second to-be-recognized image containing a key target; inputting all second to-be-recognized images corresponding to each key target into the target classification model to obtain scores or labels of the second to-be-recognized images for the multiple target types; and determining the target type of the key target according to the scores or labels of all the second to-be-identified images containing the same key target. According to the technical scheme, the targets in the to-be-identified video can be extracted, and accurate classification of key targets is realized.

Description

technical field [0001] The present invention relates to the technical field of machine vision, and more particularly, to a method and device for identifying a target type, a server, and a computer-readable storage medium. Background technique [0002] With the emphasis on public safety and the wide application of monitoring equipment, it is very important to monitor and analyze key areas such as entrances and exits or intersections. Deploy monitoring equipment in key areas such as entrances, exits or intersections to collect statistics on the types and numbers of passing objects, such as people, motor vehicles or non-motor vehicles, analyze the flow of people or vehicles in key areas, and formulate management and control of key areas. Rules are very important. [0003] In the prior art, when the type identification of the target passing through the key area is performed, the type identification is usually performed directly on the image frame extracted from the surveillance...

AI Technical Summary

Problems solved by technology

[0004] However, when the existing technology recognizes the type of the target, when the target moving speed in the image frame is fast or the target is far away, it is easy to cause the characteristics of the target to be confused, resulting in inaccurate classification results.

Images