基于改进的VGG13网络的番茄叶片病害识别

Tomato leaf disease identification based on improved VGG13 network

针对当前农作物病害的卷积神经网络识别率低等问题,文章构建了Bilinea-VGG13模型,以番茄叶片病害细粒度图片作为实验数据集,用来测试文章网络的识别性能。将VGG13网络为基础,在连接层之前加入局部池化层和双线性池化(Bi1inear Poo1),能够降低网络维度,减少参数和计算量,并且使网络能降低图片背景信息的干扰,充分提取细粒度特征,提升模型的准确率。改进后模型的准确率为98.07%,明显优于常规的卷积神经网络。为了进一步衡量网络的性能,利用测试集得出混淆矩阵计算出精确度、召回率和平均交互比(MIoU),分别为97.52%、96.44%和96.65%。实验结果表明,该方法能够有效提升模型的分类识别能力。同时,为了验证文章网络的迁移性,分别分类识别土豆、玉米和苹果的病害叶片图像,得到较高的准确率,为识别分类病害图像提供一种新的思路。

Aiming at the low recognition rate of current crop diseases by convolutional neural net work.this paper constructs Bilinear-VGG13 model.and uses fine grained images of tomato leaf diseases as experimental data set to test the recognition performance of this paper s network.Based on the VGG13 network,the Local Pooling I.ayer and Bilinear Pool are added before the connection layer,which can reduce the net work dimension,reduce parameters and calculation amount,and enable the network to reduce the interference of image background information,fully extract fine-grained features,and improve the accuracy of the model.The accuracy of the improved model is 98.07%,which is obviously better than the conventional convolutional neural network.In order to further measure the performance of the network,the accuracy,recall rate and average interaction ratio(MloU)calculated by using the confusion matrix obtained from the test set are 97.52%,96.44%and 96.65%,respectively.The experimental results show that this method can effectively improve the classification and recognition ability of the model.At the same time,in arder to verify the mobility of the network in this paper.the disease leaf images of potatoes.corn and apple were classified and identified respectively.and a higher accuracy was obtained,providing a new idea for the recognition and classification of disease images.