RT Journal
T1 OPTIMIZED DEEP BELIEF NETWORK FOR MULTI-DISEASE CLASSIFICATION AND SEVERITY ASSESSMENT IN STRAWBERRIES
A1 Venkatesh R
A1 Vijayalakshmi K
A1 Geetha M
A1 Bhuvanesh A
JF Journal of Animal and Plant Sciences
JO JAPS
SN 1018-7081
VO 35
IS 2
SP 482
OP 497
YR 2025
FD 2025/04/28
DO DOI https://doi.org/10.36899/JAPS.2025.2.0040
AB <p><span lang="EN-IN">Anthracnose, powdery mildew and gray mold are among the most destructive diseases affecting strawberries (<em>Fragaria ananassa</em>&nbsp;Duchesne), capable of rapidly spreading to healthy plants and causing significant global yield losses. Early detection of these diseases is challenging due to their f ast progression, yet it is crucial for improving strawberry yield and productivity. To address this issue, an optimal clustering-based deep learning model is proposed for the segmentation and classification of strawberry diseases using computer aided design. Initially, the input images undergo pre-processing, and the diseased areas are segmented using optimal fuzzy c means clustering with optimal centers selected by the sand cat swarm algorithm</span>. Finally, features were extracted and classified using the deep belief network model. The proposed analysis on a&nbsp;<span lang="EN-IN">strawberry disease detection dataset&nbsp;</span>demonstrated a high classification accuracy of 98.8%. The results show that the integration of optimal clustering with the deep belief network model effectively enhances classification accuracy, enabling early and accurate identification of disease lesion.</p>
K1 Computer aided design, Deep learning, Optimal clustering, Segmentation, Strawberry disease
PB Pakistan Agricultural Scientists Forum
LK https://thejaps.org.pk/AbstractView.aspx?mid=2024-JAPS-2088