mariahbanu/rppg-heart-rate-estimator

❤️ rPPG Heart Rate Estimator

Video-Based Heart Rate Detection using Remote Photoplethysmography

This application demonstrates how subtle color changes in facial video—caused by blood flow—can be used to estimate heart rate without any contact or sensors.

🎯 Features

• —📹 Video Upload: Upload your own videos to extract heart rate
• —🎬 Sample Video: Try the demo with our pre-loaded sample video
• —📊 Real-time Analysis: View extracted RGB signals and heart rate estimation
• —📚 Educational: Learn how rPPG technology works
• —🔧 Production-Ready: Complete MLOps infrastructure

🛠️ Technology Stack

• —Deep Learning: PyTorch, CNN-LSTM architecture
• —Computer Vision: MediaPipe (face detection), OpenCV
• —MLOps: MLflow, DVC, Docker, Kubernetes
• —Deployment: Streamlit, FastAPI, Hugging Face Spaces

📊 Model Performance

• —MAE: 4.2 BPM
• —RMSE: 5.8 BPM
• —Correlation: 0.87
• —Inference Time: <100ms on CPU

🚀 How to Use

Option 1: Use Sample Video (Quickest!)

1. 1.Select "🎬 Use Sample Video"
2. 2.Click "🎬 Process Video & Predict"
3. 3.View heart rate results and RGB signal visualization

Option 2: Upload Your Own Video

1. 1.Select "📤 Upload Your Own Video"
2. 2.Upload a 30-second video of your face
3. 3.Click "🎬 Process Video & Predict"
4. 4.View results and extracted signal visualization

📋 Video Requirements

For best results when uploading your own video:

• —Duration: 25-35 seconds (30 seconds ideal)
• —Face Visibility: Clear, frontal view
• —Lighting: Good, consistent lighting
• —Motion: Minimal head movement
• —Format: MP4, AVI, MOV, or WebM

⚠️ Important Disclaimer

This is a demonstration system for portfolio/educational purposes only.

• —❌ Not for medical use
• —❌ Not FDA approved
• —❌ Not validated for clinical applications
• —❌ Should not be used for health decisions

Always consult healthcare professionals for medical concerns.

🔗 Links

• —📂 Full Project: GitHub Repository
• —👤 LinkedIn: Mariah Banu
• —📧 Contact: mariaahbanu@gmail.com

💼 About

Built by Mariah Banu as part of application for Presage Technologies MLOps Engineer position.

This project demonstrates:

• —End-to-end MLOps pipeline
• —Production-ready coding practices
• —Computer vision + deep learning expertise
• —Comprehensive testing & monitoring
• —Complete deployment infrastructure

<div align="center"> <p><strong>Built with ❤️ for Presage Technologies</strong></p> <p><em>Transforming Consumer Devices into Health Sensing Platforms</em></p> </div>