摘要
松材线虫能够引起毁灭性的森林病害,松材线虫病监测技术是松材线虫病防控成功与否的关键。为了加强对森林疫情的监管力度,文中提出了一种基于神经网络的松材线虫病树统计算法。算法采用VGG16神经网络模型训练出无人机航拍图像与图像中松材线虫病树数量的映射,同时采用对称分割技术,使用病树的位置信息优化模型损失函数。实验统计结果表明,相比较不切割,适当的切割图像可使算法能够更加准确地统计病树数量。
Bursaphelenchus xylophilus can cause devastating forest diseases.And the monitoring technology of bursaphelenchus xylophilus is the key to the success of control of bursaphelenchus xylophilus.In order to strengthen the supervision of forest epidemic situation,this paper proposes a pine wilt tree statistical algorithm based on neural network.VGG16 neural network model is used to train the mapping between aerial image of UAV and the number of pine wilt trees in the image.At the same time,the symmetrical segmentation technology is used to optimize the loss function of the model by using the location information of the disease tree.At the end of the experiment,it can be concluded that the algorithm can more accurately count the number of disease trees with appropriate image cutting.
作者
雷皓辰
LEI Haochen(Wuhan Research Institute of Posts and Telecommunications,Wuhan 430073,China)
出处
《电子设计工程》
2022年第1期75-79,共5页
Electronic Design Engineering
关键词
机器学习
神经网络
计算机目标计数
松材线虫病
machine learning
neural network
computer target count
bursaphelenchus xylophilus
