Enhanced Leaf Disease Segmentation Using U-Net Architecture for Precision Agriculture: A Deep Learning Approach.

Area of Science:

• Computer Vision
• Plant Pathology
• Machine Learning

Background:

• Leaf diseases significantly impact crop yields.
• Accurate and early disease detection is crucial for effective management.
• Traditional methods often lack precision and scalability.

Purpose of the Study:

• To develop a deep learning model for precise leaf disease segmentation.
• To evaluate the U-Net architecture for identifying diseased leaf tissue.
• To enhance agricultural practices through automated disease identification.

Main Methods:

• Utilized the U-Net convolutional neural network (CNN) architecture.
• Trained and validated the model on a "Leaf Disease Segmentation" dataset with annotated images.
• Employed image preprocessing, augmentation, and regularization techniques.
• Used the Adam optimizer with a learning rate of 0.001.

Main Results:

• Achieved 99.70% training accuracy and 98.99% validation accuracy.
• Demonstrated high precision in segmenting diseased leaf regions at the pixel level.
• Showcased strong generalization capabilities on unseen data.
• Outperformed traditional image processing techniques.

Conclusions:

• The U-Net model offers a robust and accurate solution for leaf disease segmentation.
• Deep learning approaches are highly effective for plant disease identification.
• This technology has significant potential for real-world applications in precision agriculture.