基于YOLOv5s算法的汽车零件防错装检测系统

An Anti-mis-assembly Detection System for Automobile Parts Based on YOLOv5s Algorithm

在汽车生产装配过程中,如何避免汽车零件在生产线装配时发生错漏装风险是生产过程的难题。研发了一种基于YOLOv5s算法的零件防错装检测系统以解决零件错装问题。当汽车的轮毂、前后杠、尾标等零件在安装后,通过实时提取生产线摄像头视频帧,运用YOLOv5s算法识别图片中的零件,输出零件的标识(PR号),将识别出PR号与实际生产需求PR号对比,达到汽车零件防错装效果。在轮毂零件的模型训练中,所提检测系统的精确率为98.964%,召回率为97.863%,优于人工检测。

In the process of automobile production and assembly,how to avoid the risk of error or missing assembly of automobile parts in the production line is a difficult problem in the production process.A mis-assembly detection system for automobile parts based on YOLOv5s algorithm is developed to solve the problem of parts mis-assembly.After the installation of the wheel hub,front and rear bumper,tail marker and other parts of the car,the video frame from the production line camera is extracted in real time,the YOLOv5s algorithm is used to identify the parts in the picture,output the part identification(Prim?ren Number,PR number),then the PR number identified is compared with the actual PR number on the bill of material,so as to achieve the effect of preventing the mis-assembly of the car parts.In the model training of wheel hub parts,the accuracy rate of the proposed detection system is 98.964%,and the recall rate is 97.863%,which is better than manual detection.