Intelligent grading of sugarcane leaf disease severity by integrating physiological traits with the SSA-XGBoost...

Area of Science:

• Agricultural Science
• Plant Pathology
• Computational Biology

Background:

• Accurate assessment of sugarcane leaf disease severity is vital for effective crop management and yield preservation.
• Existing methods may lack efficiency or accessibility for rapid, on-site diagnosis.

Purpose of the Study:

• To develop an intelligent method for identifying sugarcane foliar disease severity using physiological traits.
• To compare the performance of various machine learning models optimized with the Sparrow Search Algorithm (SSA).

Main Methods:

• Collected physiological data (SPAD values, leaf temperature, nitrogen content) from sugarcane leaves with three diseases at four severity levels.
• Developed and optimized six classification models (KNN, AdaBoost, RF, LR, DT, XGBoost) using SSA for hyperparameter tuning.
• Normalized data using min-max scaling for model input.

Main Results:

• The Sparrow Search Algorithm (SSA) significantly enhanced the classification performance of all tested models.
• The SSA-XGBoost model demonstrated superior performance, achieving Precision, Recall, F1 Score, and Accuracy exceeding 0.9186 (PRFA score of 0.9326).
• Validation on an independent dataset confirmed strong generalization ability with 0.91 overall accuracy.

Conclusions:

• The proposed physiological trait-based method offers advantages in data accessibility, computational efficiency, and model transparency over image-based approaches.
• This study provides a reliable technical framework for intelligent diagnosis and early warning systems for sugarcane diseases.
• The findings support the practical application of this method for rapid, on-site agricultural diagnostics.