Hybrid IGWO-Dingo optimized DeMoHybridNet model for multi-class leaf disease identification

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Accurate plant disease detection is crucial for efficient crop management.
• Automated systems are needed to identify diseases in various crops like Corn, Apple, Citrus, and Mango.

Purpose of the Study:

• To propose DeMoHybridNet, an automated system for classifying crop diseases.
• To enhance classification accuracy through feature fusion and hyperparameter optimization.

Main Methods:

• Image augmentation and resizing.
• Feature extraction using DenseNet-201 and MobileNetV2.
• Feature compression via bottleneck layers and concatenation.
• Classification using Random Forest (RF).
• Hyperparameter optimization using Improved Grey Wolf Optimization-Dingo Optimization Algorithm (IGWO-DOA).

Main Results:

• DeMoHybridNet achieved high classification accuracies: 98.56% for Corn, 98.99% for Apple, 97.83% for Citrus, and 99.35% for Mango.
• The IGWO-DOA optimization improved model convergence and generalization.
• Statistical analysis confirmed the model's robustness and reliability.

Conclusions:

• The proposed optimized DeMoHybridNet is effective for automated crop disease classification.
• This technology supports precision agriculture by enabling early and accurate disease detection.
• The system demonstrates significant potential for improving crop yield and management strategies.