基于YOLOv5网络的气田无人值守场站多路入侵目标检测

Multi-Channel Intrusion Detection Method Based on YOLOv5 Network for Unattended Gas Field Station

为了优化气田无人值守场站监控效果,改善低分辨率画面检测精度低、识别困难及深度学习模型在多摄像头下资源消耗严重的问题,提出一种基于YOLOv5网络的多路入侵目标检测方法。应用YOLOv5网络及Deep SORT算法分别提取目标外观及其运动特征,通过拼接画面的方式实现对显存资源的合理利用。实验结果表明,进行迁移学习后模型的mAP值可达95%,检测精度较高,模型鲁棒性良好。

In order to optimize the monitoring effect of unattended gas field stations and improve the detection accuracy of low resolution pictures,difficulty in identification,and the serious resource consumption of deep learning model in multi-camera environment,a multi-channel intrusion target detection method based on YOLOv5 network is proposed.YOLOv5 network and Deep SORT algorithm are used to extract the appearance and motion features of the target respectively,and the reasonable utilization of the video memory resources by splicing pictures is realized.The experimental results show that the mean Average Precision(mAP)of the model after transfer learning is 95%,which proves that the detection accuracy is high,and the robustness of the model is good.