Efficient deep learning for precision viticulture: a comparative analysis of FasterViT and ViT for grape leaf disease detection
IET Conference Proceedings · 01.10.2026 · 2026 (15) · 37-42
DOI: 10.1049/icp.2026.3758 ↗Kurz gefasst
Fachartikel in der Zeitschrift IET Conference Proceedings, erschienen am 01.10.2026. Verfasst von Ankit Khandelwal und 2 weiteren Autor:innen.
Abstract (Originalsprache)
For efficient vineyard management and optimal crop yield, grape leaf disease detection and categorization are necessary. In this study, two cutting-edge deep learning architectures for multi-class grape leaf disease classification are compared: FasterViT and Vision Transformer (ViT), a pure transformer-based model. FasterViT trained on a dataset of 2,800 pictures from four classes (Leaf Blight, Black Measles, Rot, and Healthy) for 6 minutes and 19 seconds. It received a 100% test accuracy. It took ViT 11 minutes and 42 seconds to train, yet it scored a perfect test accuracy of 100% with 17.61 GMac FLOPs and 3.08 million parameters. The results indicate how to employ AI diagnostics in farming areas with little resources by exhibiting the trade-offs between accuracy, training time, and computing economy.
Zitieren
APA
Khandelwal, A., Kaur, M., & Naruka, A. S. (2026). Efficient deep learning for precision viticulture: a comparative analysis of FasterViT and ViT for grape leaf disease detection. IET Conference Proceedings, 2026(15), 37-42. https://doi.org/10.1049/icp.2026.3758
Vancouver
Khandelwal A, Kaur M, Naruka AS. Efficient deep learning for precision viticulture: a comparative analysis of FasterViT and ViT for grape leaf disease detection. IET Conference Proceedings. 2026;2026(15):37-42. doi:10.1049/icp.2026.3758
BibTeX
@article{Khandelwal2026,
author = {Khandelwal, Ankit and Kaur, Manpreet and Naruka, Aaditya S.},
title = {Efficient deep learning for precision viticulture: a comparative analysis of FasterViT and ViT for grape leaf disease detection},
journal = {IET Conference Proceedings},
year = {2026},
volume = {2026},
number = {15},
pages = {37--42},
doi = {10.1049/icp.2026.3758}
}
EUVIMED