With the continuous improvement of the intelligent level of industrial produc-tion,intelligent video recognition technology is gradually widely used in industrial produc-tion.Aiming at the complex process and difficult material tracking problems in the finishing area of Masteel's heavy H-beam steel production line,a method of material tracking is pro-posed by spraying the billet with a coding machine after the leveler to print the batch num-ber,and installing a set of character recognition device in the position of the manual check-ing table in the stacking area to recognize the batch number.At the same time,according to the characteristics of relatively small and no size or shape change of the printed charac-ters,the Yolo v3 model structure is improved,five convolutional feature maps with different scales were designed and fused with the corresponding scales in the residual network,so as to form the final feature pyramid to perform the character recognition task.The test results show that the recognition accuracy of this method can reach more than 99.5%and can be applied in practice.
关键词
重型H型钢/物料跟踪/字符识别/Yolo/v3
Key words
heavy H-beam/material tracking/character recognition/Yolo v3
万方数据