Tomato Leaf Disease Detection and Classification with Severity Estimation using Convolutional Neural Networks

Agriculture is undoubtedly the backbone of our nation. India is the second-largest producer of agricultural products globally. There are plenty of problems that a farmer faces in their field of work. When plants and crops are affected by pests, weeds, heat stress, cold stress, soil salinity, and acidity stress, drought, and flood, it affects the country's agricultural production. Usually, farmers observe the plants by their eye for detection and identification of disease. But this method can be time processing, expensive, human resources and inaccurate, etc. This proposed work is concerned with a new approach to detect the tomato plant diseases recognition model based on leaf image classification using the deep Convolutional Neural Networks (CNN) method. Automatic detection using image processing with deep learning techniques provides fast and accurate results. This technology helps the farmer identify what type of diseases the plant is being affected and suggests some medicine to be given to the affected plant. CNN techniques classified a few types of tomato diseases such as septoria leaf spot, leaf mold, bacterial spot, early blight, late blight are trained and fine-tuned to fit accurately into a plant's leaves gathered independently for diverse plant diseases. Disease severity is estimated using the dataset of infected leaf samples and output images to show the three types of severity level stages are low, medium or high levels in the MATLAB software.