Overview / Usage

Figuring out what your heart rate only requires a finger placed on your neck or wrist, a watch, and the ability to count. But if that seems like too much effort there are alternatives that can detect a heart rate using software and a camera.
Physiological data can be estimated this way thanks to the optical absorption characteristics of (oxy-) hemoglobin (see http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-16-26-21434).

Methodology / Approach

The reason this works is due to Eulerian Video Magnification, which monitors movement we can’t detect just by looking at someone. It was originally developed by a team of scientists at MIT’s Computer Science and Artificial Intelligence Lab and has a number of applications beyond heart rate detection.
Its based on research paper here http://people.csail.mit.edu/mrub/papers/vidmag.pdf

Technologies Used

The system works with just about any video footage, so I am implementing it using OPENVINO - OpenCV and Intel optimized tensorflow that will work with any webcam. My application will detect the user’s face and then focuses in on the forehead region. The software measures the average optical intensity in order to estimate the heart rate. As long as the user’s head doesn’t move too much and the lighting is good, it only takes about seconds.

Repository

https://github.com/Ujjwal-9/Heartrate-Monitor

Other links

• Research Paper