General retrieval network model for multi-class plant leaf diseases based on hashing.

Area of Science:

• Agricultural Science
• Computer Science
• Plant Pathology

Background:

• Traditional plant disease diagnosis is labor-intensive and time-consuming.
• Accurate and efficient disease identification is crucial for crop management and food security.

Purpose of the Study:

• To develop an intelligent system for enhanced plant disease retrieval and localization.
• To improve the accuracy and efficiency of disease identification in crops.

Main Methods:

• Utilized deep hash convolutional neural networks (DHCNN) for image analysis.
• Integrated a collision-resistant hashing technique to distinguish similar disease features.
• Validated the approach on single-plant and multi-plant disease retrieval scenarios, including the augmented PlantVillage dataset.

Main Results:

• Achieved over 98.4% precision and true positive rate (TPR) for single-plant disease retrieval (apple, corn, tomato).
• Attained 99.5% precision, 99.6% TPR, and 99.58% F-score for multi-plant disease retrieval.
• Demonstrated robust performance in distinguishing highly similar disease features.

Conclusions:

• The DHCNN method offers a precise and efficient solution for plant disease retrieval.
• This intelligent approach significantly reduces the need for human resources and time in disease diagnosis.
• The system proves effective in diverse and challenging plant disease identification scenarios.