Computer vision allows you to see the pulse of a person, even if he wears a mask
A year ago, the laboratory of computer science and artificial intelligence of the Massachusetts Institute of Technology developed a technology that allows you to measure a person’s pulse from a video . The algorithm enhances the smallest changes in complexion associated with the influx and outflow of blood during each heart beat. Now scientists have advanced much further - the new algorithm allows to measure the pulse, even if the video is very noisy, a person is standing with his back to the camera or wears a mask.
The new version of the video amplification algorithm does not use color changes, but micro-movements of the head caused by jolts of blood moving through the arteries. The brain is one of the main “clients” of the circulatory system, and with each pulse of the pulse, enough blood flows into it so that the head can barely sway - these movements are completely involuntary and well distinguishable after amplification. The accuracy of measuring the pulse is comparable to an electrocardiograph and even allows you to diagnose some abnormalities in the work of the heart.
')
Interestingly, before the advent of electrocardiography - the modern "gold standard" in the diagnosis of heart disease, a method called " ballistocardiography " was used in medicine, also based on the patient's micromovements. Pulse characteristics were read using instruments resembling seismographs — oscillations of the whole body or its parts were transmitted to a special sensitive platform, and then amplified. This video illustrates the principle of the ballistocardiograph:
Computer vision revives this method at a completely new level - now you can monitor the human pulse without touching it at all and without using any special equipment. This can be very useful for self-diagnosis, remote medical consultations and monitoring of patients in hospitals.
The article with a detailed description of the method can be downloaded from the link on the project page .
The new version of the video amplification algorithm does not use color changes, but micro-movements of the head caused by jolts of blood moving through the arteries. The brain is one of the main “clients” of the circulatory system, and with each pulse of the pulse, enough blood flows into it so that the head can barely sway - these movements are completely involuntary and well distinguishable after amplification. The accuracy of measuring the pulse is comparable to an electrocardiograph and even allows you to diagnose some abnormalities in the work of the heart.
')
Interestingly, before the advent of electrocardiography - the modern "gold standard" in the diagnosis of heart disease, a method called " ballistocardiography " was used in medicine, also based on the patient's micromovements. Pulse characteristics were read using instruments resembling seismographs — oscillations of the whole body or its parts were transmitted to a special sensitive platform, and then amplified. This video illustrates the principle of the ballistocardiograph:
Computer vision revives this method at a completely new level - now you can monitor the human pulse without touching it at all and without using any special equipment. This can be very useful for self-diagnosis, remote medical consultations and monitoring of patients in hospitals.
The article with a detailed description of the method can be downloaded from the link on the project page .
Source: https://habr.com/ru/post/184610/
All Articles