VGG-EffAttnNet: Hybrid Deep Learning Model for Automated Chili Plant Disease Classification Using VGG16 and...

Area of Science:

• Agricultural Science
• Computer Science
• Machine Learning

Background:

• Chili plant diseases pose significant threats to global agriculture.
• Accurate and rapid disease classification is crucial for effective crop management and yield preservation.

Purpose of the Study:

• To develop and evaluate a novel hybrid deep learning model, VGG-EffAttnNet, for robust chili plant disease classification.
• To improve the accuracy and reliability of automated disease detection systems in agriculture.

Main Methods:

• A hybrid deep learning architecture (VGG-EffAttnNet) combining VGG16 and EfficientNetB0 with attention mechanisms and Monte Carlo Dropout (MCD) was proposed.
• Extensive data augmentation was applied to a dataset of 5000 chili plant images across five classes (Healthy, Leaf Curl, Leaf Spot, Whitefly, Yellowish).
• Feature extraction, attention-based refinement, and MCD for uncertainty estimation were employed to enhance classification performance.

Main Results:

• The VGG-EffAttnNet model achieved a classification accuracy of 99%, with 99% precision and recall, outperforming individual models and state-of-the-art approaches.
• The F1-score reached 99% across all disease categories, demonstrating high classification efficacy.
• The hybrid model showed superior robustness and accuracy compared to InceptionV3 (98.83%) and MobileNet (97.18%).

Conclusions:

• The proposed VGG-EffAttnNet model offers a highly accurate and robust solution for automated chili plant disease classification.
• This deep learning approach holds significant potential for advancing precision agriculture through early disease intervention and reduced chemical reliance.
• Future research will focus on real-time deployment, explainability, and federated learning for decentralized agricultural diagnostics.