无人机智能图像识别算法在安监业务中的应用研究  

Research on application of UAV intelligent image recognition algorithm in safety supervision

作　　者：翁进荣 吴国武 魏承亮 麦照和 王思潮 姚法宇 WENG Jinrong;WU Guowu;WEI Chengliang;MAI Zhaohe;WANG Sichao;YAO Fayu(Hainan Power Grid Co.,Ltd.,Haikou 570100,China)

机构地区：[1]海南电网有限责任公司,海口570100

出　　处：《计算机应用文摘》2025年第7期90-92,95,共4页

摘　　要：为提高安全生产监督检查效率,文章提出一种基于改进深度学习的无人机智能图像识别算法。通过设计图像预处理增强方法、改进深度学习检测网络结构以及进行算法性能验证,有效解决了无人机安监过程中图像抖动、目标尺度差异大、识别实时性要求高等问题。实验结果表明,该算法在40m以下、垂直视角条件下的识别准确率达到93.2%,在边缘计算平台上可实现14.7FPS的处理速度。To improve the efficiency of safety production supervision and inspection,this article proposes a drone intelligent image recognition algorithm based on improved deep learning.By designing image preprocessing enhancement methods,improving deep learning detection network structures,and verifying algorithm performance,the problems of image jitter,large target scale differences,and high real-time recognition requirements in unmanned aerial vehicle safety monitoring processes have been effectively solved.The experimental results show that the recognition accuracy of the algorithm can reach 93.2% under the condition of vertical viewing angle below 40 m,and the processing speed of 14.7 FPS can be achieved on the edge computing platform.

关 键 词：无人机 智能图像识别 安全监管 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]