Abstract—Tomatoes represent a globally significant and commercially valuable crop, yet they are susceptible to a multitude of diseases that can significantly reduce their production and quality. To address this critical issue, we have introduced the BR-TomatoCNN, a novel lightweight Convolutional Neural Network (CNN) model that uses Bottleneck Residuals (BR) to increase the classification accuracy of tomato diseases. This research includes a comprehensive examination of how various optimizers influence the proposed model’s performance using evaluation metrics such as accuracy, loss, precision, recall, and F1-Score. A dataset consisting of nine distinct tomato disease classes collected from the Plant Village repository and the Powdery Mildew disease class was prepared with the help of farmers and experts. That was used to train the proposed model achieved remarkable results of 99.82% accuracy and an F1-Score of 1.00. These findings not only underscore the BR-TomatoCNN’s capability to accurately identify tomato diseases but also position it as a superior alternative to existing methodologies and pre-trained models. Our study underscores the significance of exploring a new approach, such as utilizing bottleneck residuals to improve the accuracy of the classification model. BR-TomatoCNN promises to play a pivotal role in disease management in the agricultural sector by facilitating early disease detection. This advancement in technology has the potential to enhance tomato crop yields and overall produce quality.
 
Keywords—convolution neural network, bottleneck residuals, image classification, plant disease detection


Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.