Apple leaf disease image recognition based on a modified rime optimization algorithm and ConvNeXt network.

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Early apple leaf disease diagnosis is crucial for crop health and agricultural productivity.
• Conventional methods struggle with complex patterns, class imbalance, and real-world challenges like poor lighting.

Purpose of the Study:

• To develop a novel model for accurate and early diagnosis of apple leaf diseases.
• To enhance feature extraction and model generalization for improved diagnostic performance.

Main Methods:

• Integration of a ConvNeXt model with the Convolutional Block Attention Module (CBAM) for feature extraction.
• Utilizing a modified rime optimization algorithm (MRIME) for hyperparameter tuning to avoid overfitting.
• Evaluation on the Apple Leaf Disease Symptoms Dataset.

Main Results:

• The proposed model achieved high performance metrics: 92.7% accuracy, 92.5% precision, 92.6% recall, 92.5% F1-score, and 92.3% mAP.
• Ablation studies showed CBAM improved accuracy by 1.5% and MRIME by an additional 1.2%.
• The model surpassed baseline models like ResNet50 and EfficientNet-B0.

Conclusions:

• The combined approach of attention mechanisms (CBAM) and metaheuristic optimization (MRIME) significantly enhances apple leaf disease detection.
• This novel model offers a state-of-the-art solution for automated plant disease diagnosis.