Aiming at the problems of poor precision and low efficiency of traditional container number localization

an improved depth neural network based on YOLOv3 algorithm is proposed to realize rapid positioning of container number. After preprocessing the collected container images

the initial priori box size required by network training is obtained through clustering

and according to the characteristics of container number positioning

the output of the network model and the loss function of network training are simplified to achieve more efficient and accurate container number positioning. Experimental results show that the positioning method of container number based on the improved YOLOv3 algorithm has high accuracy and strong real-time performance

and the positioning accuracy is as high as 98. 5 %

and the positioning speed of 26. 23 fps. The overall real-time performance and accuracy can meet the needs of practical application.