基于迁移学习的苹果树叶片病虫害识别方法研究*

为实现苹果树叶片病虫害快速且准确地识别与分类,研究基于迁移学习的多种神经网络模型,对比不同模型在苹果树叶片病虫害识别上的准确度。构建VGG16,ResNet50,Inception V3三种神经网络模型,利用从PlantVillage上获取的4种不同的苹果树叶片图片,分别为苹果黑星病叶片,苹果黑腐病叶片,苹果锈病叶片以及健康苹果叶片。按照8∶1∶1的比例将图片分为训练集,测试集和验证集对模型进行训练。在相同的试验条件下对比分析VGG16,ResNet50和Inception V3的试验结果。三种模型在使用迁移学习技术的情况下对苹果树叶片病虫害识别准确率分别达到97.67%,95.34%和100%。与不使用迁移学习的模型相比,使用迁移学习能够明显提升模型的收敛速度以及准确率,为常见的苹果树病虫害识别提供了新的方法。

Research on identification method of apple leaf diseases based on transfer learning

In order to realize the rapid and accurate identification and classification of apple leaf diseases, a variety of neural network models based on transfer learning were studied, and the accuracy of different models in apple leaf diseases identification was compared. VGG16, ResNet50, and Inception V3 neural network models are constructed. Four different apple leaf pictures were obtained from the PlantVillage, which are apple scab leaves, apple black rot leaves, apple rust leaves, and healthy apple leaves. The pictures are divided into a training set, a test set, and a verification set to train the model with a ratio of 8∶1∶1. The test results of VGG16, ResNet50, and Inception V3 were compared and analyzed under the same test conditions. The test accuracy of the three models in identifying apple leaf diseases using transfer learning technology reached 97.67%, 95.34%, and 100%, respectively. Compared with the model without transfer learning, transfer learning can significantly improve the convergence speed and accuracy of the model and provide a new method for common apple tree diseases.