Coconut Disease Detection using Traditional ML Classifiers with GLCM and SIFT Features

Coconut crops are very valuable, with both edible and non-edible parts used in many industries. These crops can suffer from diseases like bud root dropping, bud rot, grey leaf spot, yellowing of leaves, CCI caterpillars, and stem bleeding. In this paper, traditional ML models are adopted using GLCM and SIFT and combination of both. Using a custom dataset from Kaggle, we tested four ML models: RF, LR, SVM, and KNN. By blending SIFT alone and GLCM+SIFT the RF achieved perfect accuracy of 100%. LR and KNN both achieved 100% accuracy using SIFT alone, while SVM reached 92.8%. The SIFT features proved to better compared to GLCM features. This solution assist in the quick detection and classification of coconut diseases, benefiting farmers, policymakers, and agricultural stakeholders by supporting better disease management and reducing crop losses.