A Lightweight MobileViT-Based Framework for Carambola Leaf and Fruit Disease Detection

DOI

10.2991/978-94-6239-664-7_60How to use a DOI?

Keywords

Starfruit; Leaf and Fruit Disease; Deep Learning; Vision Transformer; MobileViT; Precision Agriculture

Abstract

Carambola (Averrhoa carambola) is one of the most important tropical fruits in terms of economy and nutrition. Production and quality of fruit crop can be severely affected by numerous foliar and fruit diseases. In this research, we propose a complete pipeline establishing a real-world multi-organ dataset of carambola leaves and fruits, evaluate several architectures, including common deep learning models, Vision Transformer-based architectures, and hybrid Transformer-CNN models, train an efficient, mobile-friendly model deployable in field conditions. 2,618 images were acquired in various environments and pre-processed with background cropping, normalization and extensive augmentation for the task of generalization. Multiple model families were trained and tested, and the best performing MobileViT hybrid architecture was able to achieve an overall accuracy of 99.67% along with comparable precision, recall, F1 score performance using less than 0.95 million parameters only. Grad-CAM interpretability analysis further demonstrated that the model successfully highlighted disease-related regions, improving reliability and interpretability. Additionally, the pre-trained model was also implemented into an Android application for on-device disease diagnosis without relying on high-end computers. The major part of this research contributes to precision agriculture by reducing reliance on expert inspection and enabling timely intervention to minimize crop loss while supporting sustainable, data-driven cultivation.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - S. M. Abdullah Al Muhib
AU  - Rejowan Arifin Nayeem
AU  - Shalim Shadman Eshan
AU  - Jarin Tasnim Showrin
AU  - Anisa Khatun Bristy
AU  - Shahriar Marjan
AU  - Nafiz Ahmed Emon
PY  - 2026
DA  - 2026/06/08
TI  - A Lightweight MobileViT-Based Framework for Carambola Leaf and Fruit Disease Detection
BT  - Proceedings of the International Conference on Intelligent Data Analysis and Applications (IDAA 2025)
PB  - Atlantis Press
SP  - 872
EP  - 887
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-664-7_60
DO  - 10.2991/978-94-6239-664-7_60
ID  - AlMuhib2026
ER  -