[4Xin1] Poster session 2

[4Xin1-34] Countermeasures and verification of sudden infant death syndrome using object detection.

〇Yuji HASHIMOTO¹, Kenichi TERANISHI¹ (1.yzlogic)

Keywords:SIDS, Anomaly Detection, Object Detection

Sudden Infant Death Syndrome (SIDS) is a disease that causes infants with no warning signs or medical history to die. In 2021, 81 people in Japan died from SIDS. Although the cause is unknown, it is believed that sleeping face down or with a blanket covering the face prevents heat from escaping from the face and causes heat to accumulate in the body. Therefore, in this study, we considered a system for detecting changes in the condition of infants using a surveillance camera. The method involved preparing a doll representing an infant, training its face using the object detection algorithm Yolo, and detecting four patterns of situations in which a blanket is placed over the face. Change Finder was used to analyze the change points. As a result, changes in the probability of detection indicated changes in the infant's condition.