Non-Contact Breathing Rate Estimation Using Machine Learning with an Optimized Architecture

The breathing rate monitoring is an important measure in medical applications and daily physical activities. contact sensors have shown their effectiveness for been mostly used as a standard reference, but with some disadvantages example burns patients vulnerable skins. Contactless systems are then gaining attention respiratory frequency detection. We propose new non-contact technique to estimate the based on motion video magnification method by means of Hermite transform Artificial Hydrocarbon Network (AHN). chest movements tracked system without use ROI image video. machine learning classifies frames inhalation or exhalation using Bayesian-optimized AHN. was compared optimized Convolutional Neural (CNN). This proposal has tested Data-Set containing ten healthy subjects four positions. percentage error Bland–Altman analysis compare performance strategies estimating rate. Besides, search agreement estimation reference.The AHN 2.19±2.1 respect reference ≈99%.