Neural Network Heart Rate Detection via Facial Video PPG
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing heart rate measurement technologies face challenges such as the risk of pathogen transmission, time-consuming clinic visits, and inaccurate readings in dynamic or poorly lit conditions, especially when using camera-based systems.
Innovation Solution
A neural network-based method that uses a camera to obtain a video clip of a subject's face, detects facial features, converts image color space to L*a*b*, and predicts a photoplethysmographic signal using a deep neural network, followed by Fourier transform analysis to determine heart rate, enabling remote and accurate measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional contact-based heart rate monitoring devices are used, then measurement can be obtained, but there is risk of pathogen transmission and requires clinic visits
Solution Approach 1:
The patent uses video imaging as an intermediary to capture heart rate information remotely. Instead of direct contact with the subject, the system captures video of the subject's face or body, extracts physiological signals through image processing, and determines heart rate without physical contact, thereby eliminating pathogen transmission risk while maintaining measurement reliability
Solution Approach 2:
The patent replaces mechanical contact-based sensing (such as pulse palpation or electrode contact) with optical-based video imaging. By substituting the mechanical measurement system with an optical system that captures blood flow-induced color changes in video frames, the system achieves contactless heart rate monitoring that eliminates the need for physical contact and associated infection risks
2Ease of operation
If camera-based heart rate measurement is used, then remote measurement is enabled, but accuracy deteriorates in challenging conditions such as head movement or poor lighting
Solution Approach 1:
The patent applies dynamic image processing techniques that adapt to changing conditions. The system uses motion compensation algorithms to track and stabilize the region of interest despite head movements, and employs adaptive filtering to maintain signal quality under varying lighting conditions. This dynamic approach allows the system to maintain measurement precision while preserving the ease of remote operation
Solution Approach 2:
The patent performs preliminary processing steps including video stabilization, region of interest extraction, and preliminary signal filtering before final heart rate calculation. By preparing and pre-processing the video data in advance, the system compensates for challenging conditions such as head movement and poor lighting, ensuring accurate measurements are achieved even in suboptimal environments
3Measurement precision
If deep neural network processing is applied to video frames, then heart rate measurement accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent segments the deep neural network processing into distinct functional modules: video preprocessing, feature extraction networks, signal generation, and heart rate calculation. By dividing the complex processing task into smaller, specialized segments, the system achieves high measurement precision while making the overall system more manageable and efficient, allowing parallel processing and optimization of individual components
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach allows for remote, accurate, and efficient heart rate measurement in various conditions, reducing the need for physical contact and improving accuracy by mitigating the effects of head movements and lighting variations.
Implementation Method 1
the deep neural network predicts a photoplethysmographic (PPG) signal based on the sequence of images
Implementation Method 2
applying a Fourier transform to the PPG signal to convert the PPG signal to the frequency domain
Data Source
AI summary
In some examples, an electronic device comprises an interface to receive a video of a human face, a memory storing executable code, and a processor coupled to the interface and to the memory. As a result of executing the executable code, the processor is to receive the video from the interface, use a facial detection technique to produce a sequence of images of the human face based on the video, use a neural network to predict a photoplethysmographic (PPG) signal based on the sequence of images, convert the PPG signal to a frequency domain signal, and determine a heart rate by performing a frequency analysis on the frequency domain signal.