Tomato plant leaf disease segmentation and multiclass disease detection using hybrid optimization enabled deep...

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Plant diseases pose a significant threat to global food security and crop quality.
• Early detection and treatment of plant diseases are crucial for sustainable agriculture.
• Traditional methods for disease identification can be time-consuming and less accurate.

Purpose of the Study:

• To develop and evaluate a Deep Learning (DL)-assisted technique for detecting and classifying tomato plant diseases.
• To enhance the accuracy and efficiency of plant disease identification using advanced computational models.
• To improve crop yield and quality by enabling timely intervention against plant diseases.

Main Methods:

• Utilized a deep batch-normalized eLu Alex Net (DbneAlexnet) model for image classification.
• Employed anisotropic filtering for preprocessing tomato leaf images to remove distortions.
• Implemented U-net with Gradient-Golden search optimization (Gradient-GSO) for image segmentation.
• Applied image augmentation techniques, including position and color augmentation.
• Trained the DbneAlexnet model using a proposed Gradient Jaya-Golden search optimization (GJ-GSO) algorithm.

Main Results:

• The DL-assisted technique achieved a high accuracy of 92.4% in classifying tomato plant diseases.
• Demonstrated a True Positive Rate (TPR) of 91.9% and a True Negative Rate (TNR) of 92.2%.
• Achieved a minimal False Positive Rate (FPR) of 0.078, indicating high reliability.
• The unified segmentation and classification approach proved effective in identifying plant diseases.

Conclusions:

• The developed DL model, GJ-GSO-based DbneAlexnet, is an effectual tool for identifying plant diseases.
• The technique offers significant benefits for early disease detection, contributing to improved crop management.
• Empirical research validates the effectiveness and accuracy of the developed model for agricultural applications.