A HYBRID OPTIMIZATION AND DEEP LEARNING APPROACH FOR CROP DISEASE DETECTION AND PESTICIDE RECOMMENDATION

Abstract

Crop diseases remain a major threat to global agriculture, food security, and economic stability, making rapid and accurate detection essential for effective management. Traditional diagnostic methods are often time-consuming, labor-intensive, and inadequate for addressing newly emerging complex diseases. To address these challenges, this study proposes a Crop Ensemble Net approach for disease detection and pesticide recommendation. The method to apply advanced preprocessing techniques to improve the resolution of images and identify distinguishing features between diseased and healthy leaves. Feature selection and optimization are further improved using Mud Ring Optimization (MRO) and Secretary Bird Optimization (SBO), which strengthen detection accuracy. The proposed ensemble integrates three high-performing deep learning architectures—Efficient Net, ResNet, and InceptionV3—to balance speed, precision, and multi-scale feature extraction. Efficient Net ensures computational efficiency, ResNet captures intricate hierarchical structures, and InceptionV3 identifies disease patterns across varying scales. Experimental results demonstrate that the Crop Ensemble Net achieves superior accuracy and robustness compared to existing methods, offering a scalable and interpretable solution for sustainable crop disease management.