ベビーマット型デバイスを用いた重心・重量計測に基づく乳児の啼泣検出及び予測手法の検討

In recent years, supporting families during the childcare period has become increasingly important amid declining birthrates. In particular, responding to a sudden infant cry can lead to accumulated fatigue and increased anxiety among caregivers. Therefore, technologies that can detect the occurrence of crying, capture its early signs, and notify caregivers are needed. Conventional approaches have mainly relied on audio or video data; however, these modalities can be affected by background noise in daily living environments and raise privacy concerns. In this study, we propose a method for detecting and predicting infant crying using a baby-mat-type device equipped with load cells. The device measures oscillations in the infant’s center of gravity and weight caused by body movement on the mat. We collected center-of-gravity and weight data from 18 infants aged five months or younger and evaluated the proposed method using data from seven infants with sufficient crying samples. The results achieved average F1 scores of 0.887 and 0.741 for cry detection and cry prediction, respectively. These findings indicate the feasibility of detecting and predicting infant crying based on body-movement-related center-of-gravity and weight oscillations without relying on audio or video data.