Laparoscopic surgery is an active area of research in clinical practice. Detecting tools in surgical videos can help physicians manipulate surgical instruments, reduce complications and ensure patient safety. In this paper, in order to improve the accuracy of laparoscopic surgical instruments detection and due to the problems of low accuracy, large volume and inconvenient deployment of the detection model, we propose a laparoscopic surgical instruments detection system based on YOLOv5 model deployed on the edge AI equipment produced by EMA for real-time detection of laparoscopic surgical tools. Localization and identification covered seven commonly used surgical tools. The experimental results show that the system has good performance on the Cholec80 laparoscopic surgical instrument detection dataset with a mean accuracy (mAP) of 98.8% and a inference speed of more than 37 frames per second (fps) on the edge AI equipment.