arkul07/RescueLens

Disaster Response & Low-Resource Medical Triage MVP RescueLens is a lightweight, AI-powered medical triage system designed for disaster response scenarios and low-resource medical environments. It uses browser-side computer vision with MediaPipe Pose detection and real-time WebSocket communication to provide instant patient assessments. # 🚑 RescueLens - AI-Assisted Triage System **Disaster Response & Low-Resource Medical Triage MVP** RescueLens is a lightweight, browser-based AI-powered medical triage system designed for disaster response scenarios and low-resource medical environments. It uses browser-side computer vision with MediaPipe Pose detection and real-time WebSocket communication to provide instant patient assessment and triage decisions. ## 🌟 Features ### 🎥 Browser-Side Video Analysis - **MediaPipe Pose Detection**: Lightweight, real-time pose detection in the browser - **Breathing Pattern Analysis**: Sliding window FFT-based respiratory rate estimation - **Movement Classification**: Purposeful vs. low vs. none movement detection - **Multi-Patient Tracking**: Simultaneous detection and tracking of multiple patients - **Visual Overlays**: Real-time triage status with color-coded patient boxes ### 🎤 Audio Processing (Optional) - **Web Audio API**: Real-time microphone analysis for breathing detection - **Keyword Detection**: Simple keyword recognition for distress calls - **Audio Quality Assessment**: SNR-based audio signal quality evaluation ### 🧠 START-Like Triage Rules - **Automated Triage Decisions**: RED, YELLOW, GREEN, BLACK, UNKNOWN classification - **Confidence Scoring**: Multi-factor confidence calculation - **Human Override Support**: Manual override capability with reason tracking - **Real-Time Updates**: Live triage status via WebSocket ### 📊 Data Management - **Event Logging**: Comprehensive logging of AI decisions and human overrides - **Export Functionality**: JSON and CSV export of triage logs - **Privacy-First**: No raw video/audio data sent to backend ## 🏗️ Architecture ### Frontend (React + Vite + TypeScript) - **Browser-Side Processing**: MediaPipe Pose detection runs locally - **Real-Time Analysis**: Sliding window breathing and movement analysis - **WebSocket Communication**: Live data streaming to backend - **Modern UI**: Clean, responsive interface optimize …