An NLP-Based Approach for Tomato Leaf Disease Prediction and Classification

Abstract

Early identification of tomato leaf diseases is essential to minimize crop loss and improve agricultural productivity. Traditional disease diagnosis methods rely heavily on manual inspection, which is time-consuming, subjective, and requires expert knowledge. To overcome these limitations, this work presents an NLP-based intelligent approach combined with deep learning for tomato leaf disease prediction and classification. The proposed system utilizes a structured workflow that begins with dataset acquisition followed by preprocessing, feature extraction, and classification. A pre-trained MobileNetV2 deep learning model is employed for effective feature learning and disease recognition. The model is trained using labeled tomato leaf images representing multiple disease categories such as early blight, late blight, leaf mold, bacterial spot, and target spot. The extracted features are analyzed through deep neural layers to generate accurate predictions. Experimental results demonstrate that the proposed system achieves 75% classification accuracy, with a precision of 81.15% and recall of 70.33%, indicating reliable disease detection performance. The model successfully identifies disease types and generates detailed diagnostic reports, including disease name, confidence score, symptoms, severity level, and recommended treatment measures. Visual performance analysis using accuracy and loss curves confirms stable convergence and effective learning behaviour. The developed system offers a practical, automated, and scalable solution for tomato disease diagnosis, reducing dependency on manual inspection and expert intervention. The integration of deep learning with structured output reporting makes the system suitable for real-world agricultural applications and future deployment in smart farming environments.