Research on automatic billet turning control system based on BiSeNet

That the turning of steel billets in the rolling process is completed manually by operators affects the improvement of rolling automation level. A solution based on machine vision for billet edge identification and billet turning control in billet production is proposed. Firstly, the real-time billet image is obtained through image acquisition platform. Secondly, a semantic segmentation model based on bilateral semantic convolutional neural network (BiSeNet) is developed to identify the billet edge. Finally, the automatic control system is designed to realize the automatic turning of billet. The utilization rate of the system is up to 90%, and the application of the control system can reduce the workload of operators and improve the automation level of steel plants.