Categorizing Plant Leaf Disease using Convolutional Neural Network

Plant disease could lead to remarkable losses in agricultural production, which also affect yield and quality. Implementing effective control measures for these diseases requires accurate and prompt identification. With a focus on the late blight and early blight of plant leaves, the researchers in this paper suggest a framework for categorizing plant diseases. Researchers have employed transfer learning strategies and Convolutional Neural Networks (CNNs) to construct a trustworthy and efficient disease categorization model. The suggested method uses a customized sequential model with well-known CNN algorithms for training and testing, such as ResNet-50, VGG16, VGG19, and the Sequential Model. With the pre-trained weights from these models, transfer learning is used to efficiently obtain features from the provided data. During training, a 30% dropout rate is used to improve model generalization and reduce overfitting. To evaluate the effectiveness of our method, the authors have acquired a substantial amount of data from potato leaves, including samples with early and late blight diseases. Experimental results depict that our suggested strategy using a sequential model has outperformed all other transfer learning algorithms and classifies potato leaf diseases with a good accuracy rate of 98.7% on testing data with 50 epochs, a batch size of 32, and a loss of 0.04. On the other hand we were able to reach an accuracy of 96%, recall score of 0.9624, precision score of 0. 9594, and f1 score of 0.9608 when we employed the ResNet model. Such a system can be beneficial to farmers and agricultural specialists make quick judgements and put effective disease management tactics into practice, ultimately improving crop health and yield.