基于Faster RCNN的间作类农田无人机喷雾方法研究  

Study on UAV spray method of intercropping farmland based on faster RCNN

作　　者：王宜全 吴扬东[1] 祁兵 Wang Yiquan;Wu Yangdong;Qi Bing(Key Laboratory of Advanced Manufacturing Technology of Ministry of Education Guizhou University ,Guiyang550025 ,China;Guizhou Eagle Brother UAV Innovation Co. Ltd. ,Guiyang550001,China)

机构地区：[1]贵州大学现代制造技术教育部重点实验室,贵阳市550025 [2]贵州天鹰兄弟无人机科技创新有限公司,贵阳市550001

出　　处：《中国农机化学报》2019年第6期76-81,共6页Journal of Chinese Agricultural Mechanization

基　　金：贵州省科技支撑项目(黔科合支撑[2016]2037)——植保无人机静电喷雾系统研发

摘　　要：为解决植保无人机在间作类农田进行农药喷洒作业时对非喷雾对象的误喷问题.提出一种基于Faster RCNN目标检测的无人机喷雾方法.包括图像预处理、模型训练与测试、喷雾规则的制定。对模型识别效果与喷雾规则可行性进行试验,结果表明,该识别模型对喷雾对象的识别准确率达96.66%,对喷雾对象位置定位的准确率达91.33%,该喷雾规则具有较高的可行性。研究结果为间作类农田的无人机智能化喷雾提供参考。In order to solve the problem of mis-sprayingto the non-spraying objects during the pesticide spraying operation of UAV on the intercropping farmland. A UAV spray method based on Faster RCNN target detection is proposed, including image preprocess ing ,model training and testing, and formulation of spray rules. Experiments on the model recognition effect and the feasibility of the spray rules are conducted. The results show that the recognition accuracy of the trained model for the spray object is 96. 66%, and the accuracy of the position of the spray object is 91. 33%. The spray rule has high feasibility. A UAV intelligent spray method for inter cropping farmland is provided by the results of the study.

关 键 词：间作类农田 喷雾对象 FASTER RCNN 喷雾规则 

分 类 号：S126[农业科学—农业基础科学]