live-stream-object-detection

Share your webcam stream with real-time object detection to the public by Node.js.

Technologies

This project using the following technologies to go:

• opencv4nodejs: Use it to capture webcam frames and more power
• TensorFlow Object Detection: Use it to detect objects on a frame
• FFmpeg: Use it to compress frames and make the live streaming
• ws: Use it to send stream to web browser
• JSMpeg: Use it to decode streaming from server for web browser
• ngrok: Use it to get public and secure URL for exposing the local web server
• express: Use it to make a web server

Requirements

• A webcam, e.g. laptop-integrated webcam, USB webcam
• Node.js 8 or higher
• OpenCV 3.3.1 or higher
• FFmpeg 3.3 or higher

Installation

The below commands to install the project.

1. npm install
2. node server.js

Running the demo

• Make sure you are still in the project_root directory
• To run the server: node server.js then you will get the Ngrok HTTP URL in terminal
• To run the demo, open a browser and go to Ngrok HTTP URL

The app should be up and running!