Automated Orange Fruit Disease Detection and Classification Using an Enhanced VGG-MHA-LSTM Deep Learning Model
VFAST Transactions on Software Engineering · 20.06.2026 · 14 (2) · 281-297
DOI: 10.21015/vtse.v14i2.2464 ↗Kurz gefasst
Fachartikel in der Zeitschrift VFAST Transactions on Software Engineering, online erschienen am 20.06.2026. Verfasst von C. Mahalakshmi und 1 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben.
Abstract (Originalsprache)
Early detection of orange diseases is essential for maintaining fruit quality and reducing economic losses. Surface lesions and blotches caused by diseases lower the market value of oranges and increase post-harvest decay. Traditional visual inspection methods are time-consuming, subjective, and impractical at large scale, making automated detection necessary. This study proposes an automated disease identification and classification system for oranges using a VGG-MHA-LSTM model, which combines VGG-16 for deep feature extraction with a Long Short-Term Memory (LSTM) network enhanced by Multi-Head Attention (MHA). Before feature extraction, images are preprocessed using a Wiener filter for noise removal and Contrast Limited Adaptive Histogram Equalization (CLAHE) for contrast enhancement. RGB-thresholding based color segmentation is then applied to isolate diseased regions from the fruit surface. The extracted deep features are passed into the MHA-LSTM classifier to identify the most discriminative features for disease classification. The proposed model was evaluated on a publicly available orange dataset from the Kaggle repository. The results show that the VGG-MHA-LSTM model achieved a classification accuracy of 98.95%, demonstrating its effectiveness in distinguishing between citrus canker, melanose, and healthy orange classes. The proposed approach provides a reliable and accurate method for automated orange disease classification and has practical potential for integration into agricultural inspection workflows to support more precise disease diagnosis.
Zitieren
APA
Mahalakshmi, C., & Saravanan, S. (2026). Automated Orange Fruit Disease Detection and Classification Using an Enhanced VGG-MHA-LSTM Deep Learning Model. VFAST Transactions on Software Engineering, 14(2), 281-297. https://doi.org/10.21015/vtse.v14i2.2464
Vancouver
Mahalakshmi C, Saravanan S. Automated Orange Fruit Disease Detection and Classification Using an Enhanced VGG-MHA-LSTM Deep Learning Model. VFAST Transactions on Software Engineering. 2026;14(2):281-297. doi:10.21015/vtse.v14i2.2464
BibTeX
@article{Mahalakshmi2026,
author = {Mahalakshmi, C. and Saravanan, S.},
title = {Automated Orange Fruit Disease Detection and Classification Using an Enhanced VGG-MHA-LSTM Deep Learning Model},
journal = {VFAST Transactions on Software Engineering},
year = {2026},
volume = {14},
number = {2},
pages = {281--297},
doi = {10.21015/vtse.v14i2.2464}
}
EUVIMED