Design and Implementation of a Deep Learning–Based Multimodal Childcare Support System for Deaf Parents

Recent advances in machine learning and Internet of Things (IoT) technologies have led to the development of childcare support services that monitor infants’ conditions in real time and provide immediate feedback. However, most existing services are designed for non-disabled users and rely heavily on auditory cues, which creates significant limitations for deaf parents who have difficulty perceiving sound-based situations. Although several prior studies have proposed childcare support systems for deaf parents, technical limitations that fail to adequately reflect real-life environments remain. To address these issues, this study designs and implements a deep learning–based multimodal childcare support system. The proposed system consists of three core functions: infant cry detection and cause classification, hazardous sound detection, and infant suffocation risk detection. An event-driven, hardware-based notification framework is employed to deliver alerts to a smartwatch. System demonstrations confirm stable real-time operation of all functions and successful alert transmission when hazardous situations occur.