摘要
为了实现对储粮害虫的有效防治,发展一种正确高效的害虫识别方法是非常重要的。在计算机视觉与深度学习技术的协助下,通过深度卷积神经网络建模进行害虫图像识别。在该方法下实现了储粮害虫特征的自动提取,并在基于图像的7个类别的储粮害虫识别上取得了98.81%的识别正确率。因此,基于深度学习的储粮害虫特征提取与分类方法具有很高的实用价值,可以进一步推广到实际的储粮害虫防治和粮仓管理中来。
Developing a correct and efficient stored-grain pest recognition method is very important when confronted with the problem of preventing and controlling the damage caused by stored-grain pests.With the help of computer vision and deep learning technology,we can use deep convolutional neural networks to process the images of stored-grain pests.The method could automatically extract the features of pests images and it also achieved an accuracy rate up to 98.81% on the image dataset of 7 different classes of pests.Therefore,the method of feature extraction and image recognition based on deep learning is very valuable in practical use and can be further extended to the practical stored-grain pests control and granary management.
出处
《皖西学院学报》
2017年第5期67-72,共6页
Journal of West Anhui University
基金
安徽省自然科学基金项目"基于基因表达式编程的作物生长建模方法研究"(1508085MF110)
茶树生物学与资源利用国家重点实验室开放基金(SKLTOF20150103)
关键词
储粮害虫
特征提取
图像识别
卷积神经网络
深度学习
stored-grain pests
feature extraction
image recognition
convolutional neural networks
deep learning
