YOLOv9-Based Detection of Diseases in Poplar Trees Using Histogram Equalization and Computer Vision.

Area of Science:

• Plant Pathology
• Computer Vision
• Forestry Science

Background:

• Poplar trees are vital for industry and ecosystems, necessitating effective disease management.
• Convolutional Neural Networks (CNNs) offer advanced capabilities for plant disease detection and classification.
• Accurate identification of poplar diseases is crucial for sustainable forestry and resource management.

Purpose of the Study:

• To create and release the first publicly available, geographically diverse dataset of diseased poplar leaves.
• To develop and evaluate a robust system for detecting and classifying common poplar diseases using deep learning.
• To enhance the accuracy of poplar disease diagnosis through image preprocessing techniques.

Main Methods:

• Collection of a diverse poplar leaf image dataset from Uzbekistan and South Korea, covering four disease classes: Parsha (Scab), Brown spotting, White-Gray spotting, and Rust.
• Application of Histogram Equalization for image preprocessing to improve visual quality and aid disease detection.
• Utilization of the YOLOv9c Convolutional Neural Network (CNN) model for disease classification and detection.

Main Results:

• The developed system achieved high accuracy in detecting and classifying four common poplar diseases.
• The Histogram Equalization preprocessing step significantly enhanced the performance of the YOLOv9c model.
• The creation of a novel, publicly accessible dataset for diseased poplar leaves.

Conclusions:

• The study provides a valuable, publicly available resource for advancing research in poplar disease detection.
• The integration of YOLOv9c and Histogram Equalization offers a scalable and accurate solution for monitoring poplar tree health.
• This work contributes to sustainable forestry practices through improved early disease detection and management strategies.