Transforming Agriculture: A Vision Transformer Approach for Tomato Disease Detection

Mudaliar Saurabh Ravi, K. S. Archana, Ritik Ranjan Sinha, Harsh Kumar Singh

2025

Abstract

Tomato diseases significantly affect agricultural yield, making automated classification crucial. Deep learning models, particularly Convolutional Neural Networks (CNNs) like Efficient Net, have achieved high accuracy in this domain. However, Vision Transformers (ViTs) present a novel alternative by leveraging self-attention for feature extraction. This study compares google/vit-base-patch16-224, pretrained on ImageNet via Hugging Face, with Efficient Net to assess their effectiveness. The ViT model attained a training accuracy of 84%, while Efficient Net outperformed it with 95% accuracy. Despite this, ViT demonstrated superior generalization and interpretability. These findings underscore the trade-offs between CNNs and Transformers, highlighting ViT’s potential for scalable and explainable disease detection in agriculture. Future research will explore hybrid models and dataset augmentation to enhance ViT’s performance.

Paper Citation

in Harvard Style

Ravi M., Archana K., Sinha R. and Singh H. (2025). Transforming Agriculture: A Vision Transformer Approach for Tomato Disease Detection. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 833-839. DOI: 10.5220/0013890600004919

in Bibtex Style

@conference{icrdicct`2525,
author={Mudaliar Ravi and K. Archana and Ritik Sinha and Harsh Singh},
title={Transforming Agriculture: A Vision Transformer Approach for Tomato Disease Detection},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={833-839},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013890600004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Transforming Agriculture: A Vision Transformer Approach for Tomato Disease Detection
SN - 978-989-758-777-1
AU - Ravi M.
AU - Archana K.
AU - Sinha R.
AU - Singh H.
PY - 2025
SP - 833
EP - 839
DO - 10.5220/0013890600004919
PB - SciTePress