Design of multi-target workpiece tracking system based on darknet29

Abstract

Aiming at the problem of poor tracking performance caused by slow workpiece detection in industrial automation pipeline, an intelligent multi workpiece tracking method based on target detection network is proposed. Firstly, using darknet29 feature extraction network framework, a faster computing network structure is designed as the backbone of workpiece detector; secondly, the bottleneck residual block and RFB module are optimized to compress the network parameters; thirdly, the multi-dimensional information is fused to construct the similarity matrix, and the Hungarian algorithm is used to obtain the optimal matching of multi-objective artifacts; finally, a single target tracking method is proposed to solve the problems of missed detection and false detection in multi-target tracking and improve the tracking quality. The conclusion is drawn from the comparison experiment with the mainstream multi-target tracking algorithm. While the detection and tracking algorithm ensures the accuracy and robustness on the workpiece data set, the FPS reaches 26.3 as the optimal value, and the tracking performance is superior.