<![CDATA[Citrus leaf conditions can affect crop quality and productivity if not identified accurately and promptly. Manual identification is often time-consuming and depends on the observer's ability to recognize visual symptoms. This study aims to develop an image-based citrus leaf condition classification model using a Hybrid MobileNetV3-SVM approach and compare its performance with a standalone MobileNetV3-Small model. The dataset consisted of eleven categories of citrus leaf conditions. The data were divided into training, validation, and testing sets using a ratio of 70:15:15. MobileNetV3-Small was employed as a feature extractor through transfer learning and fine-tuning, while a Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel was used as the classifier in the hybrid model. To address class imbalance, Random Under Sampling (RUS) was applied to the training data before model training. Experimental results showed that the Hybrid MobileNetV3-SVM model achieved an accuracy of 98.18%, precision of 98.20%, recall of 98.18%, and F1-score of 98.18%. In comparison, the standalone MobileNetV3-Small model achieved an accuracy of 97.27%, precision of 97.32%, recall of 97.27%, and F1-score of 97.25%. The results indicate that incorporating SVM improved classification performance and reduced the number of misclassified samples in the testing dataset. Therefore, the Hybrid MobileNetV3-SVM approach can be utilized as an effective method for classifying citrus leaf conditions based on digital images.]]>
Copyrights © 2026
