<![CDATA[Non-contact respiratory rate measurement in infants presents an innovative alternative to traditional contact-based methods, which often lead to discomfort. This study aims to develop an automated approach for measuring infant respiratory rate via a non-contact method using video recordings. The method automatically detects the Region of Interest (ROI) in the infant's torso and estimates the respiratory rate using optical flow. A pose estimation model is employed to detect the ROI automatically. The method was developed and tested on the AIR-125 video dataset, which includes various lighting conditions, infant poses, and frame rates. Results demonstrate that the proposed method effectively detects the torso and provides reliable respiratory rate estimations with a mean absolute error of 3.82 BPM and Root Mean Square Error 5,01 BPM. This system offers a flexible, non-contact solution for monitoring infant respiratory rate suitable for both home and clinical settings.]]>
Copyrights © 2025
