Nabilah Ari Adzani, Muhammad Cahyo Bagaskoro, Aripriharta, Risfrendra, Suherman, Gwo-Jiun Horng
Stunting is one of the chronic health problems in infants that affects physical growth and cognitive development. This study developed Peduli Stunting 5.0, a system for classifying the nutritional status of infants based on image detection using YOLOv5 integrated with Internet of Things (IoT) and cloud technology. The dataset consists of 344 images of infants labeled into three categories: healthy, stunted, and malnourished. The YOLOv5 model was trained to detect and classify nutritional status based on visual body features of infants. Evaluation results show good performance in the healthy category (AUC-PR 0.963) and stunting category (AUC-PR 0.883), with an overall detection accuracy of 0.716 (mAP@0.5). Although malnutrition classification remains low (AUC-PR 0.304), the model shows potential as an early screening tool. Integration with a mobile application enables real-time detection and data recording in the field. This system can strengthen efforts to mitigate stunting, particularly in areas with limited access to healthcare services. ©2025 IEEE.
Department of Electrical Engineering and Informatics, Universitas Negeri Malang, Malang, Indonesia; Department of Electrical Engineering, Universitas Negeri Padang, Padang, Indonesia; Department of Electrical Engineering, Universitas Sumatera Utara, Medan, Indonesia; Department of Electronics Engineering, Southern Taiwan University of Science and Technology, Taiwan, Taiwan
