Automatic Diagnosis of Soybean Leaf Disease by Transfer Learning

Abstract

Soybean diseases and insect pests are important factors that affect the output and quality of soybeans, thus it is necessary to do correct inspection and diagnosis of them. For this reason, based on improved transfer learning, this study proposed a classification method for soybean leaf diseases. Firstly, leaves were segmented from the whole image after removing the complicated background images. Secondly, the data-augmented method was applied to amplify the separated leaf disease image dataset to reduce overfitting. At last, the automatically fine-tuning convolutional neural network (Autotun) was adopted to classify the soybean leaf diseases. The verification accuracy of the proposed method is 94.23, 93.51 and 94.91% on VGG-16, ResNet-34 and DenseNet-121 networks respectively. Compared with the traditional finetuning method of transfer learning, the results show that this method is better than the traditional transfer learning method.

Towards Improved Disease Identification With Pretrained Convolutional Neural Networks as Feature Extractors for Chili Leaf Images

“…The accuracy, recall, specificity, precision, and F1score are the indexes used to assess the identification performance of the utilized CNN models in the research. These indexes are computed from the confusion matrix generated according to the testing result produced by the models, which are True Positive (TP), True Negative (TN), False Positive (FP), and False Negative (FN) [26]. Accuracy is the measure of a likelihood of correct prediction made by a CNN model.…”