Deep Learning-based Tomato Leaf Disease Detection and Classification

Low productivity in agriculture is mainly caused because of the plant diseases. As the plants play major role in the food production, the disease developed in the plants must be eradicated. In recent years, Artificial Intelligent (AI) techniques, especially in the field of image processing has been creating significant impact. Those AI based image processing techniques are employed in the deep learning based algorithms to achieve early detection of the diseases in the plants. In this study, the deep learning algorithms play vital role in the detection and classification of the diseases in the tomato leaves. The Pixel Optimized Triplet Attention Hybrid CNN-BiLSTM classification model used in this study predicts the diseases in the tomato leaves more accurately. The study is also compared with the other existing algorithms such as ANN, CNN, ResNet-50, and LSTM for proving the significant work done by the proposed model. Moreover, the work of the proposed Pixel Optimized Triplet Attention Hybrid CNN-BiLSTM classifier is examined by using the evaluation metrics which has achieved 98.50% of accuracy, 98.2% of precision and 98.4% of recall. Thus the study proves the earliest detection and classification of the disease in tomato leaves can be achieved dynamically only by the use of this proposed Pixel Optimized Triplet Attention Hybrid CNN-BiLSTM classifier. Hence, by detecting the disease in the tomato leaves earlier, the food production in the agriculture is effectively increased.