Abstract
Abstract
Early and reliable detection of plant wilting is crucial for maintaining crop productivity and ensuring sustainable agricultural practices. This study aimed to evaluate the effectiveness of machine learning models in detecting wilting in
Vinca rosea
plants using simple geometric features extracted from digital images. The extracted features included leaf width, length, area, aspect ratio, and area coverage. Four machine learning classifiers, Support Vector Machine (SVM), Decision Tree (DT), k-Nearest Neighbours (kNN), and Naive Bayes (NB) were trained and validated to classify plants as wilted or healthy. Results showed that SVM achieved the highest performance, with 0.98 accuracy, 0.99 precision, 0.95 recall, 0.97 F1 Score, and 0.99 area under the curve (AUC), outperforming the other models. DT and kNN achieved competitive accuracy (0.97 and 0.95, respectively) but exhibited weaker recall, while NB showed high recall (0.92) but reduced precision (0.87). The findings highlight the robustness of SVM and the potential of simple geometric features combined with machine learning for rapid and reliable wilting detection. Moreover, integrating such image-based wilting detection with irrigation control can optimise water use while sustaining plant health and yield.