EfficientRMT-Net-An Efficient ResNet-50 and Vision Transformers Approach for Classifying Potato Plant Leaf Diseases.

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Potato diseases, such as early and late blight, significantly reduce global crop yield and quality.
• Traditional disease detection methods are labor-intensive, time-consuming, and often inaccurate.
• Automated systems are needed for efficient and precise potato leaf disease identification.

Purpose of the Study:

• To develop an advanced, automated system for early detection and classification of potato leaf diseases.
• To integrate Vision Transformer (ViT) and ResNet-50 architectures into a novel model, EfficientRMT-Net.
• To overcome limitations of existing methods by improving accuracy and efficiency in disease identification.

Main Methods:

• Developed EfficientRMT-Net, combining Vision Transformer (ViT) and ResNet-50 architectures.
• Utilized Convolutional Neural Network (CNN) for feature extraction and depth-wise convolution (DWC) for computational efficiency.
• Incorporated a stage block structure for improved scalability and sensitive area detection.
• Trained and validated the model on custom potato leaf image datasets.

Main Results:

• EfficientRMT-Net achieved 97.65% accuracy on a general image dataset.
• The model reached 99.12% accuracy on a specialized potato leaf image dataset.
• EfficientRMT-Net outperformed existing deep learning and transfer learning techniques in classification accuracy.

Conclusions:

• The EfficientRMT-Net model provides an efficient and accurate solution for classifying potato plant leaf diseases.
• The integration of ViT and ResNet-50 architectures effectively addresses complex agricultural challenges.
• This automated system can help farmers enhance crop yield and optimize resource utilization.