Automated Tomato Leaf Disease Classification Using CNNs for Real-Time Field Application

Tomato leaf diseases are a significant threat to crop yield and agricultural productivity across the world. Conventional detection methods rely on human eye inspection, which are both time-consuming and susceptible to error. In this study, we apply deep learning techniques with CNNs in order to automatically identify and classify tomato leaf diseases. The proposed CNN model was trained on a comprehensive dataset of tomato leaf images encompassing healthy leaves and those affected by various diseases. Techniques of data augmentation were also used to enhance model generalization. The CNN architecture was designed to extract relevant features from leaf images and classify them into different disease categories. This model obtains an average test accuracy of 97.31% in classifying the various diseases that affect tomato leaves to surpass some of the pre-trained models. In a case of a multiple class, the model gives robust performance with an average precision of 97.3% and average recall of 97.3%. An important contribution of this study is the ability to deploy a trained model in a mobile application for realtime field disease detection using React Native. This approach will provide good accessibility for early and accurate detection of tomato leaf disease. It will, thereby significantly reduce crop losses and improve productivity in agriculture through timely interventions.