.png){kind=icon}
.png){kind=icon}
Return
SAUFCMC-OSCN: deep learning framework with enhanced kookaburra optimization for high-precision black pepper leaf disease classification
.png){kind=icon}
PRE
AI
R
T
DOI：10.1080/02286203.2026.2680221.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Disease detection in black pepper cultivation is vital for sustaining crop yield and economic stability. This study introduces an advanced Deep Learning (DL) framework for automated diagnosis of black pepper leaf diseases. The work leverages the BPLDD06 Dataset (six disease classes, collected from Kasaragod, Kerala) along with a secondary potato leaf disease dataset (3,076 images) for comparative evaluation. The novelty lies in two core innovations: (i) the Sheaf Attention U-Net with Fuzzy C-Means Clustering (SAUFCMC), which enhances lesion boundary segmentation through attention-based spatial coherence and soft-clustering, and (ii) the Orientation-Steered Crystal Net (OSCN), which applies crystallography-inspired orientation features, enabling detection of subtle tissue disruptions often missed by CNNs and Transformers. Additionally, the Enhanced Kookaburra Optimization Algorithm (EKOA) ensures robust parameter tuning by balancing exploration and exploitation for stable convergence. The proposed framework achieves 98.81% accuracy, 98.03% precision, 98.06% recall, 98.04% F1-score, and 97.75% specificity in black pepper disease detection, and 98.76% accuracy on potato leaf disease detection. Statistical validation confirms the reliability of these results, positioning the framework as a robust and scalable solution for early agricultural disease detection.
Keywords:
Black pepper disease detection
sheaf attention UNet
enhanced Kookaburra optimization algorithm
orientation-steered crystal net
fuzzy C-means clustering
Journal
I
IF:
3.9
Papers:
596
Citations:
1.5K
Organization
Cited Papers