Automated Detection and Severity Prediction of Wheat Rust Using Cost-Effective Xception Architecture.

Area of Science:

• Agricultural Science
• Plant Pathology
• Computer Vision

Background:

• Wheat rust diseases, caused by Puccinia triticina, pose a significant threat to global wheat production.
• Current manual methods for detecting wheat rust are labor-intensive and inefficient for timely disease management.
• Existing control strategies often lack effective early identification of disease outbreaks.

Purpose of the Study:

• To develop an automated computer vision pipeline for predicting wheat leaf rust severity.
• To provide a low-cost, accessible, and efficient tool for early disease screening in field conditions.
• To enhance wheat crop management through rapid and accurate disease assessment.

Main Methods:

• A deep learning classifier was used to distinguish between healthy and rust-infected wheat leaves.
• Grabcut-based segmentation was employed to isolate infected leaf areas.
• Image processing in the CIELAB color space identified rust features, and disease severity ratio was calculated.

Main Results:

• The pipeline successfully differentiated healthy from infected wheat leaves.
• Segmentation and color space analysis effectively isolated and characterized rust symptoms.
• A quantifiable disease severity ratio was computed for infected leaves.

Conclusions:

• The proposed computer vision pipeline offers a groundbreaking, automated solution for wheat rust disease screening.
• This method provides a low-cost and accessible alternative to manual inspection for field conditions.
• The approach facilitates timely interventions and improved control measures for wheat rust, advancing crop disease management.