Hybrid feature optimized CNN for rice crop disease prediction.

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Autonomous systems are vital in agriculture for rice disease detection to prevent yield loss.
• Current disease identification methods struggle with accuracy and computational efficiency.
• Accurate leaf segmentation and disease stage analysis are critical challenges.

Purpose of the Study:

• To develop a more accurate, cost-effective, and reliable method for rice disease detection.
• To introduce a hybrid bio-inspired algorithm (Hybrid WOA_APSO) for optimizing feature selection.
• To enhance rice disease classification using Convolutional Neural Networks (CNNs).

Main Methods:

• Proposed a Hybrid WOA_APSO algorithm, merging Adaptive Particle Swarm Optimization (APSO) and Whale Optimization Algorithm (WOA).
• Utilized CNN for disease classification in rice crops.
• Conducted experiments on benchmark datasets (Plantvillage) focusing on feature extraction, segmentation, and preprocessing.

Main Results:

• The Hybrid WOA_APSO algorithm optimized feature selection for improved CNN accuracy.
• Achieved a high classification accuracy of 97.5% for rice diseases.
• Demonstrated superior performance compared to Support Vector Machine (SVM), Artificial Neural Network (ANN), and conventional CNN models.

Conclusions:

• The proposed Hybrid WOA_APSO-CNN model offers a significant advancement in automated rice disease detection.
• This approach enhances accuracy and efficiency, addressing limitations of existing methods.
• The findings provide a foundation for further research in intelligent agricultural systems.