一种基于YOLOv5检测轨道螺丝的机器人

A Robot for Detecting Rail Screws Based on YOLOv5

在列车行驶过程中,对轨道的安装和维护尤为重要,因此设计了一种基于YOLOv5检测轨道螺丝的机器人。该机器人主要利用YOLOv5算法检测轨道螺丝解决铁道上螺丝滑丝的问题。利用2000张轨道螺丝的样本进行深度神经网络训练,实现基于YOLOv5检测轨道螺丝检测与识别。该方案在轨道螺丝缺陷或者残缺数据集检测精度较高,漏检情况较少,定位能力较强,检测速度较快,符合轨道安全实时检测的要求,为轨道维护工作中的轨道螺丝检测工作提供了参考方案。

In the process of train running, it is particularly important to install and maintain the track. This paper proposes a robot based on YOLOv5 to detect the track screws. It mainly uses YOLOv5 algorithm to solve the problem that the track screws are not fixed or incomplete on the railway. At the same time, 2000 samples of track screws are used to train in-depth neural networks to realize the detection and recognition of track screws based on YOLOv5 detection. This scheme improves the detection accuracy of track screw defects or incomplete data sets, reduces missed detection, has strong positioning ability and fast detection speed. It meets the requirements of real-time detection of track safety, and provides a reference scheme for track screw detection in track maintenance.