[1]邬 昀,吕 浩,陈 希,等.基于图像识别的气象雨量筒异物检测[J].工业仪表与自动化装置,2026,(02):64-71.[doi:10.19950/j.cnki.CN61-1121/TH.2026.02.013]
　WU Yun,LV Hao,CHEN Xi,et al.Image-based detection of foreign objects in meteorological rain gauges[J].Industrial Instrumentation & Automation,2026,(02):64-71.[doi:10.19950/j.cnki.CN61-1121/TH.2026.02.013]
点击复制

基于图像识别的气象雨量筒异物检测()

《工业仪表与自动化装置》[ISSN:1000-0682/CN:61-1121/TH]

卷:
期数:
2026年02期
页码:
64-71
栏目:
出版日期:
2026-04-15

文章信息/Info

Title:
Image-based detection of foreign objects in meteorological rain gauges
文章编号:
1000-0682（2026）02-0064-08
作者:
邬 昀1; 2; 吕 浩3; 陈 希3 ; 张 旭3
1.湖北省气象信息与技术保障中心；2.临近空间环境特性及效应全国重点实验室，湖北 武汉 4300742； 3.航天新气象科技有限公司，江苏 无锡 214127
Author(s):
WU Yun1; 2; LV Hao3; CHEN Xi3; ZHANG Xu3
Hubei Meteorological Information and Technology Support Center, 2..State Key Laboratory of Environment Characteristics and Effects for Near-space,Hubei Wuhan 430074，China；3.Aerospace New Meteorological Technology Co., Ltd., Jiangsu Wuxi 214127，China
关键词:
自动气象站; 雨量筒; YOLOv5s-SE; 目标识别; 异物检测
Keywords:
automatic weather station;  rain gauge; YOLOv5s-SE; target recognition; state detection
分类号:
TP391
DOI:
10.19950/j.cnki.CN61-1121/TH.2026.02.013
文献标志码:
A
摘要:
作为气象站中测量降水量的重要工具，保证雨量筒处于正常的工作状态十分重要。为了实现野外复杂情况下对自动气象站中雨量筒被异物覆盖的实时检测，提出一种基于YOLOv5s改进的YOLOv5s-SE模型的雨量筒目标异物检测。将SE注意力机制加入YOLOv5s主干网络，增强模型特征提取的能力，降低对雨量筒特征提取时受到的干扰，实现雨量筒的快速、准确检测。结果表明，改进模型的损失值维持在0.022左右，精确度高达98.9%，模型预测的准确率在90%以上，充分证明了改进模型在复杂条件下的鲁棒性和高精度识别能力。该文提出YOLOv5s-SE对自动气象站的雨量筒异物覆盖进行识别研究，能够实现无人环境下的实时监测，达到自动化精准识别，做到异常状态及时纠正，有力保障了后续雨量筒对降水量的准确测量。
Abstract:
As an important tool for measuring precipitation in meteorological stations, it is very important to ensure that the rain gauge is in a normal working state. order to achieve real-time detection of the rain gauge in the automatic weather station being covered by foreign bodies in the complex field environment, a YOLOv5s-SE based on YOLOv5s improvement is proposed for the detection of foreign bodies covering the rain gauge target. Integrating the SE attention mechanism into the backbone network of YOLOv5s to enhance the model’s ability to extract features, reduce interference during the extraction of features for rain gauge detection, and achieve fast and accurate detection of rain gauges.The results show that the loss value of the improved model maintains around 0.022, the accuracy is as high 98.9%, and the accuracy of the model prediction is above 90%, which fully proves the robustness and high-precision recognition ability of the improved model complex conditions. This paper proposes YOLOv5s-SE to recognize and study the foreign body coverage of the rain gauge in the automatic weather station, which can achieve realtime monitoring in an unmanned environment, reaching automated and accurate recognition, and making timely corrections to abnormal States, which effectively guarantees the accurate measurement of precipitation by the rain gauge behind.

参考文献/References:

[1] 孙姝芹,孙新龙.综合气象观测业务质量控制综述[J]. 南方农业, 2018,12(24): 159-160.

[2] 范雯杰,陈立峰,熊鸣,等. 基于LoRa的自动气象站故障诊断装置的设计与实现[J]. 计算机测量与控制,2024,32(07): 7-14+29.
[3] 王飞.新型区域自动气象站的故障判断及处理[J].河南科技, 2020, (10): 145-147.
[4] 黄春娟,徐安伦,杨根铨,等.云南复杂地形下区域自动气象站的故障分析[J].气象水文海洋仪器, 2024,41(06): 125-129.
[5] 宋树生,胡颖飞,崔日权.新型自动气象站故障处理及日常维护管理[J].农业灾害研究,2020,10(07): 28-29+31.
[6] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2014:580-587
[7] Girshick R.Fast R-CNN[C]//2015 IEEE International Conference on Computer Vision (ICCV).Santiago:IEEE,2015:1440-1448.
[8] Ren S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[9] REDMON J, DIVVALA S, GIRSHICK R, et al. You Only Look Once:Unified, Real-Time Object Detection[C]//Proceedings of the 29th IEEE Conference on Computer Vision and Pattern Recognition, 2016:779-788.
[10] REDMON J, FARHADI A. YOLO9000:better, faster, stronger[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, 2017:7263-7271.
[11] REDMON J, FARHADI A. YOLOv3:An Incremental Improvement[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018:89-95
[12] LIU W, ANGUELOV D, ERHAN D, et al. SSD:Single Shot Multibox Detector[C]//Proceedings of the 14th European Conference on Computer Vision. Amsterdam, 2016:21-37.
[13] 王鑫杰,王吉平.YOLO目标检测算法综述[J]. 广西物理, 2024, 45(02): 50-53.
[14] 孔凡国,仇展明,王鑫,等. 基于YOLOv5的药瓶识别定位系统设计[J].机械工程与自动化,2025,54(01): 188-189+192.
[15] 沈斌,罗晓倩,王超.基于改进YOLOv5s的煤矿车辆车牌检测方法[J].矿冶,2025,34(01):177-184.
[16] 陈江川,曹芷怡,余赠宇,等.基于YOLOv5的农田病虫害检测识别方法[J].南方农机, 2025,56(04):112-117.
[17] 王红君,刘紫宾,赵辉,等.基于改进YOLOv5的苹果轻量化检测算法[J].农机化研究, 2025, 47(07): 65-71.
[18] 张宏飞,冯永利,黄金凤.改进YOLOv5的输送带缺陷检测[J].电子测量技术,2024,47(22):161-168.
[19] 李季,李恒昶,曾晨,等.基于图像识别的雨量筒异物检测系统的设计与实现[J].自动化技术与应用,2025,44(06):149-154.
[20] 杨飞帆,李军.面向自动驾驶的YOLO目标检测算法研究综述[J].汽车工程师, 2023(11): 1-11.
[21] 戴佳兵,宋春芳,凌彩金,等. 基于YOLOv5s-SE和通道剪枝的虫咬紫金蝉茶检测方法研究[J].河南农业科学, 2024, 53(05): 157-163.
[22] 江培营,陶青川,艾梦琴.基于注意力机制和深度学习的钢板表面缺陷图像分类[J].计算机应用与软件, 2021,38(09):214-219.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期：2025-10-08基金项目：湖北省自然科学基金气象创新发展联合基金项目（2024AFD207）第一作者：邬昀（1983—），男，湖北荆门人，高级工程师，硕士，主要从事气象探测技术和装备保障智能化研究。
更新日期/Last Update: 1900-01-01