Abstract
Plant diseases have a major impact on food security and agricultural productivity because they lower crop quality and output. For crop management to be effective, plant diseases must be identified early and accurately. Using leaf image analysis, this study suggests an intelligent deep learning-based approach for automatically identifying illnesses in tomato plants (Solanum Lycopersicon). Plants that produce tomatoes are particularly vulnerable to a number of diseases, including leaf Mold, early blight, and late blight, which can negatively affect crop yield. In order to increase feature visibility, a dataset of tomato leaf photos that includes both healthy and diseased samples is gathered and preprocessed utilizing image enhancement and normalization techniques. In order to precisely classify various disease categories and extract deep information from leaf photos, the suggested method uses a Convolutional Neural Network (CNN) in conjunction with transfer learning utilizing the Mobile Net architecture. Significant visual characteristics like Colour shifts, textural alterations, and lesion shapes on infected leaves are immediately picked up by the CNN model. The suggested model's efficacy is shown by experimental study. Strong classification capabilities are demonstrated by the system's 97.8% training accuracy and 96.4% testing accuracy. The algorithm's reliability is further supported by performance assessment metrics, which show an F1-score of 96.0%, recall of 96.1%, and precision of 95%. Stable model convergence is also shown by the training and validation loss values, which drop from 0.45 and 0.50 at the first epoch to 0.11 and 0.15, correspondingly. The suggested MobileNet-CNN models works better than current designs like VGG16 (92.3%), ResNet50 (94.1%), and InceptionV3 (95.2%), according to comparison studies. By offering a quick and accurate tool for early disease identification, the proposed system can assist cultivators and farming specialists, facilitate prompt medical care and enhance total crop production and methods that are environmentally friendly.