Tackling unbalanced datasets for yellow and brown rust detection in wheat

Area of Science:

• Agricultural Science
• Remote Sensing
• Computer Science

Background:

• Wheat rust diseases pose significant threats to global food security.
• Early and accurate detection of plant diseases is crucial for effective crop management.
• Hyperspectral imaging offers detailed spectral information for vegetation analysis.

Purpose of the Study:

• To evaluate the efficacy of hyperspectral data for detecting yellow and brown rust in wheat.
• To compare the performance of various machine learning models in rust detection.
• To assess the impact of the Synthetic Minority Oversampling Technique (SMOTE) on model accuracy.

Main Methods:

• Utilized hyperspectral data collected from wheat crops.
• Applied machine learning algorithms: Artificial Neural Network (ANN), Support Vector Machine (SVM), Random Forest (RF), and Gaussian Naïve Bayes (GNB).
• Employed SMOTE to address imbalanced datasets during model training.

Main Results:

• Support Vector Machine (SVM) and Random Forest (RF) models demonstrated superior performance.
• RF achieved 70% accuracy for yellow rust detection without SMOTE.
• SVM achieved 63% accuracy for brown rust detection with SMOTE applied.

Conclusions:

• Hyperspectral data and machine learning techniques show significant potential for wheat rust detection.
• SMOTE can enhance model performance, particularly for imbalanced disease datasets.
• Further research into data processing and augmentation techniques is recommended for optimizing plant disease detection systems.