摘要
图像识别是除草机器人的一项基础关键研究。为了能提高农作物和杂草的识别率以及便于识别物特征的提取,提出了基于卷积神经网络的识别方法。以农田中的杂草和农作物为试验对象设计了网络结构。该网络结构的参数较少,准确率达到了92.08%,且处理每张图片的时间仅为0.82ms。
Image recognition is the basic and key research of weeding robot.In order to improve the recognition rate of crops and weeds and facilitate the extraction of recognition features,we proposed a recognition method based on convolutional neural network.The network structure was designed with weeds and crops as experimental objects.The network structure had fewer parameters with its accuracy reaching 92.08%,and the processing time of each picture was only 0.82 ms.
作者
张有春
ZHANG You-chun(School of Computer Science and Technology,Henan Polytechnic University,Jiaozuo,Henan 454000)
出处
《安徽农业科学》
CAS
2019年第14期242-244,共3页
Journal of Anhui Agricultural Sciences
关键词
除草
卷积神经网络
快速
图像识别
Weeding
Convolutional neural network
Fast
Image identification