Potato plant disease detection: leveraging hybrid deep learning models

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Crop diseases pose a significant threat to global food production and economic stability.
• Accurate and early detection of potato plant diseases is crucial for maintaining yield and productivity.
• Existing detection methods, including traditional and some AI approaches, struggle with accuracy and real-world variability.

Purpose of the Study:

• To develop and evaluate a novel hybrid deep-learning model for enhanced potato plant disease detection and identification.
• To improve the accuracy and generalizability of disease detection in complex agricultural environments.
• To address the limitations of current methods in identifying potato diseases under variable field conditions.

Main Methods:

• A hybrid deep-learning model, EfficientNetV2B3+ViT, was developed by integrating a Convolutional Neural Network (EfficientNetV2B3) with a Vision Transformer (ViT).
• The model was trained on the diverse "Potato Leaf Disease Dataset", encompassing images representative of real-world agricultural conditions.
• Performance was evaluated based on accuracy in disease detection and identification.

Main Results:

• The proposed EfficientNetV2B3+ViT hybrid model achieved a detection accuracy of 85.06%.
• This represents an 11.43% improvement in accuracy compared to previous study results.
• The model demonstrated effectiveness in identifying potato diseases within complex agricultural settings.

Conclusions:

• The hybrid EfficientNetV2B3+ViT model offers a significant advancement in potato plant disease detection and identification.
• This approach shows strong potential for practical application in agriculture, enhancing disease management strategies.
• The study underscores the efficacy of hybrid deep learning models for improving crop health monitoring.