<![CDATA[Depression is a mental health disorder that can affect an individual's emotional condition, behavior, and quality of life. Facial expressions can represent a person's emotional state and therefore have the potential to be utilized as a visual source of information in the development of artificial intelligence-based classification systems. This study aims to develop a depression indication classification model based on facial expressions using MobileNetV2 as a feature extractor and Support Vector Machine (SVM) as a classifier. The FER2013 dataset was used and grouped into two classes, namely depression indication and non-depression indication based on predefined facial expression categories used in this study. After the labeling process, a total of 19,275 facial images were obtained, with 3,855 images used as testing data. The proposed method consists of image preprocessing, feature extraction using MobileNetV2, classification using SVM, threshold optimization, and model evaluation. Experimental results show that the proposed model achieved an accuracy of 79.69% with an AUC value of 88.62%. Threshold optimization produced an optimal threshold value of 0.44 and improved the accuracy to 80.34%. The precision, recall, and F1-score values indicate relatively balanced performance across both classes. The results demonstrate that the combination of MobileNetV2 and SVM can provide good classification performance on the FER2013 dataset grouped into depression indication and non-depression indication classes.]]>
