摘要
本系统采用YOLOv5实现有害生物检测算法,手机端采用NCNN部署模型,离线数据库采用SQLite,同时实现WEB端平台,完成系统管理与可视化统计分析。通过真实环境下的测试得出,mAP值约为93%,测试精度约为71%。在安卓设备上进行识别约耗时100ms,较好地实现了功能。
This system uses YOLOv5 to implement pest detection algorithms,the mobile phone uses the NCNN deployment model,the offline database uses SQLite,and the WEB platform is implemented at the same time to complete system management and visual statistical analysis.According to the test under the real environment,the mAP value is about 93%,and the test accuracy is about 71%.It takes about 100ms to recognize on the Android device,and the function is well realized.
作者
周为鹏
方定磊
顾仕杰
颜新云
Zhou Weipeng;Fang Dinglei;Gu Shijie;Yan Xinyun(Jinling Institute of Technology Network and Communication Engineering,Nanjing Jiangsu,211100;Nanjing Agricultural University Artificial Intelligence Academy,Nanjing Jiangsu,210018;Hangzhou Dianzi University IT Academy,Hangzhou Zhejiang,310018)
出处
《电子测试》
2022年第15期56-59,共4页
Electronic Test
