<![CDATA[Stunting is a chronic nutritional problem that will directly affect the quality of human resources in the future. One of the contributing factors to stunting is anemia during pregnancy, which often originates from adolescence. Early detection of anemia in women of reproductive age is a crucial preventive measure to reduce the risk of stunting. This study aims to develop an anemia classification model based on conjunctival images using a combination of MobileNetV2 architecture and Support Vector Machine (SVM), and to implement the model into a mobile application named CeSLA (Cegah Stunting Lewat Anemia). The model was built using a dataset of female conjunctival images annotated based on haemoglobin levels and visual characteristics of the conjunctiva. Evaluation results explain that the model achieved precision, recall, and f1-score values ranging from 0.91 to 0.92 for each class, with a macro average of 0.92, indicating accurate and balanced classification performance. The trained and evaluated model was then integrated into the CeSLA mobile application. This application allows users, particularly adolescent girls, to detect potential anemia non-invasively by scanning the lower eyelid using a smartphone camera. CeSLA is also equipped with educational features such as health articles and a detection history log. With this approach, CeSLA is expected to serve as an innovative solution that supports early, self-administered anemia detection and contributes to the national effort to prevent stunting.]]>
