pyVHR

pyVHR estimates heart rate variability from video recordings using remote photoplethysmography (rPPG) by extracting blood volume pulse (BVP) signals to compute beats per minute (BPM) in realistic environments.

Key Features:

• Multi-Stage Pipeline: Implements a comprehensive multi-stage pipeline for extracting and analyzing heart rate fluctuations from video data to derive BVP and BPM.
• Support for Various Methods: Supports development, assessment, and statistical analysis of traditional and learning-based rPPG methods across multiple datasets.
• Accelerated Processing Capabilities: Uses accelerated Python libraries for video and signal processing and parallel/accelerated ad-hoc procedures to enable online processing on a GPU with an average speedup of around five times for 30 fps HD videos.
• Real-Time Processing: Provides mechanisms for safe and efficient real-time operation on video inputs.

Scientific Applications:

• Large-Scale Health Monitoring: Enables contactless monitoring of heart rate variability for population-scale or non-contact screening scenarios.
• Telemedicine and Remote Patient Care: Facilitates remote acquisition of BVP and BPM from video for telehealth assessments and follow-up.
• Continuous Monitoring in Dynamic Environments: Supports continuous heart rate monitoring from high-definition video in environments with realistic motion and lighting variability.
• rPPG Method Development and Validation: Serves as a platform for comparing, developing, and statistically evaluating traditional and learning-based rPPG algorithms.

Methodology:

Computational methods include a multi-stage pipeline for extracting and analyzing BVP from video, accelerated Python video and signal processing libraries, parallel/accelerated ad-hoc procedures for online GPU processing, and statistical analysis supporting traditional and learning-based rPPG methods.