基于卷积神经网络的黄瓜病害识别研究

Research on Cucumber Disease Recognition Based on Convolution Neural Network

针对在黄瓜种植过程中,不能及时观察出病害种类以及不合理地使用药物防治而导致减产或死亡的问题,提出了基于卷积神经网络的黄瓜病害识别方法。通过使用手机拍照的方法采集带有病害特征的样本图片,进行图像增强处理,制作了黄瓜叶面病害数据集,并研究AlexNet、VGG-16和ResNet50三种不同深度网络模型的病害识别效果,通过设计不同方案进行模型训练,找出训练效果最优的网络模型并进行病害图片检测。结果表明,系统能够满足预期的黄瓜病害识别要求,具有较高的识别准确率。

In order to solve the problem that the variety of cucumber diseases can not be observed in time and the irrational use of drugs leads to the reduction of production or death,a cucumber disease identification method based on convolutional neural network is proposed.Sample pictures with disease characteristics were collected by taking pictures with mobile phones,and then image enhancement was carried out to make cucumber leaf disease data sets.The disease recognition effects of three different depth network models,AlexNet,VGG-16 and ResNet50,were studied.By designing different schemes for model training,the network model with the best training effect was found and the disease pictures were detected.The results showed that the system could meet the expected requirements of cucumber disease recognition and had high recognition accuracy.