Fall Prediction and Prevention: Smart Onesie

Fall hazards are a prevalent and leading cause of death in babies and toddlers nowadays. It is found that mothers and parents who work confront the majority of these challenges. Most of the time, newborns go unnoticed when in childcare or under the care of nannies, which can put them in perilous or life-threatening situations. In the case of babies in a crib, whether asleep or awake, there are numerous postures such as crawling near the crib slats, standing by gripping the edge of the crib fence, climbing at the crib's gate, roll-over, and so on that can lead to a fall or serious injuries. The aim of this work is to identify those events by recognizing postures using computer vision techniques such as image mapping and the pretrained object detection algorithm. To optimize the subject (i.e., risky postures) of this work, the COCO dataset is utilized in the SSD-MobileNetv3 method. The challenges encountered in this part of the study are the primary rationale for extending the work in the future and incorporating a new MODNET-based model to identify unsafe postures.