Towards Region-Adapted Deep Learning Models for Date Palm Disease Detection in the UAE
Document Type
Conference Proceeding
Source of Publication
Lecture Notes in Networks and Systems
Publication Date
8-2-2026
Abstract
The date palm (Phoenix dactylifera L.) holds significant economic and cultural value across the Gulf region, particularly in the United Arab Emirates (UAE). However, date palm cultivation is increasingly threatened by several diseases that can reduce productivity and potentially impact regional food security. Therefore, early identification and diagnosis of these diseases are essential to enable rapid intervention, limit crop losses, and support effective farm management. In this study, the use of deep learning for the automated classification of 12 common date palm diseases in UAE is investigated. The dataset includes 490 images collected during field visits to date palm farms in collaboration with experts from the Abu Dhabi Agriculture and Food Safety Authority (ADAFSA). Several convolutional neural network (CNN) architectures were evaluated for disease recognition. Unlike prior studies that rely on large public datasets or simplified two- to three-class problems, this work addresses a 13-class setting using field-collected images from the UAE farms. Among the tested models, DenseNet121 achieved the best performance, reaching a validation accuracy of 80.61% and a macro-averaged F1-score of 74.89%. Model performance across individual disease categories is also analyzed, and challenges commonly encountered in real agricultural datasets, particularly class imbalance and limited labeled data availability, are discussed. The collected dataset and experimental results provide an initial benchmark for applying deep learning methods to date palm disease classification using field-collected images from UAE farms.
DOI Link
ISBN
[9783032295576]
ISSN
Publisher
Springer Nature Switzerland
Volume
2036 LNNS
First Page
142
Last Page
154
Disciplines
Computer Sciences
Keywords
Computer vision, Date palm, Deep learning, Disease classification, Precision agriculture, Transfer learning
Scopus ID
Recommended Citation
Almarzooqi, Ayesha; Alhosani, Sara; Almansoori, Shamma; Grati, Rima; and Kohail, Sarah, "Towards Region-Adapted Deep Learning Models for Date Palm Disease Detection in the UAE" (2026). All Works. 8036.
https://zuscholars.zu.ac.ae/works/8036
Indexed in Scopus
yes
Open Access
no