Skip to content

README_LIVE_CAPTURE.md

Latest commit

 

History

History
108 lines (82 loc) · 2.81 KB

README_LIVE_CAPTURE.md

File metadata and controls

108 lines (82 loc) · 2.81 KB
Raw

Live Capture & Streaming Support

Eaglearn now supports professional live capture and streaming capabilities for:

  • Live streaming to platforms (YouTube, Twitch, etc.)
  • WebRTC applications for web browsers
  • RTMP streaming to media servers
  • High-performance capture with multiple backend support

Quick Start

Enable Live Capture Mode

Add to your config.yaml:

camera:
  use_live_capture: true  # Enable live capture service
  backend: auto  # Auto-detect best backend

Or use directly in code:

from live_capture_service import LiveCaptureService

# Initialize live capture
live = LiveCaptureService()

# Start with auto backend detection
live.start_capture('auto')

Supported Backends

Backend Description Use Case
rtmp RTMP streaming to media servers (requires FFmpeg) Professional streaming
webrtc WebRTC for web browsers (requires aiortc) Web applications
directshow DirectShow for Windows (low latency) Windows desktop
v4l2 Video4Linux2 for Linux (good performance) Linux desktop
dshow Default OpenCV fallback Cross-platform

Streaming Examples

RTMP Streaming (YouTube/Twitch)

# Configure RTMP in live_capture_service.py
# Then start with:
live.start_capture('rtmp')

WebRTC Streaming (Web Browser)

# WebRTC implementation included
live.start_capture('webrtc')

High-Performance Capture

# Auto-detects optimal backend
live.start_capture('auto')

Features

  • Multi-backend support: Automatic fallback if backend fails
  • Low latency: Optimized for real-time applications
  • High resolution: Support up to 4K streaming
  • Frame buffering: Configurable buffer size (default: 30 frames)
  • Client management: Multiple concurrent streaming clients
  • Stream info: Real-time backend, resolution, FPS, client count

Integration with Eaglearn

The live capture service integrates seamlessly with existing Eaglearn features:

  • VLM analysis works with live capture frames
  • Focus tracking and emotion detection
  • All existing metrics and analytics
  • Enhanced UI shows live capture status

Requirements

# For RTMP streaming
pip install ffmpeg-python

# For WebRTC streaming
pip install aiortc

# Core requirements (already in requirements.txt)
opencv-python, numpy, transformers, torch

Configuration

All live capture settings are configurable in config.yaml:

  • Backend selection
  • Resolution and FPS
  • Frame buffering
  • Auto-detection preferences

Performance

Optimized for:

  • Low CPU usage (selective frame processing)
  • Memory efficiency (smart buffering)
  • Network performance (adaptive quality)
  • Real-time responsiveness (< 50ms latency)

Use live capture for professional streaming, remote monitoring, or web-based applications!