rtm2

English | 中文

A real-time motion monitor

Table of Contents

• Introduction
• Install and usage
• Todo
• Similar projects

Introduction

rtm2 is a real-time motion monitor developed by python3, flask, socketIO and openCV. Unlike similar projects' transferring every frame to clients, rtm2 send frames and notices to clients using websocket protocol only when motion🏃 detected. It can send data real-time and avoid unnecessary traffic transfer.

There are many motion detecting projects that combined openCV with network. As far as I know, clients get frames that captured by openCV in the way of solutions shown below, mainly.

1. ajax/xhr polling
clients(browser) send query to server every N seconds:

function check_update() {
   var xhr = new XMLHttpRequest();
   xhr.open('get','/some/url/to/check/update',true);
   xhr.send();
   if(xhr.readyState === 4){
       if(xhr.status === 200){
           // function to show image
       }
   }
}
setInterval(check_update, 1000)

It's not real-time and could cause heavy server load.

2. multipart protocal
clients browser

<img src="{{ url_for('video_feed') }}">

server

def gen():
    """Video streaming generator function."""
    while True:
        read_return_code, frame = vc.read()
        encode_return_code, image_buffer = cv2.imencode('.jpg', frame)
        io_buf = io.BytesIO(image_buffer)
        yield (b'--frame\r\n'
               b'Content-Type: image/jpeg\r\n\r\n' + io_buf.read() + b'\r\n')


@app.route('/video_feed')
def video_feed():
    """Video streaming route. Put this in the src attribute of an img tag."""
    return Response(gen(), mimetype='multipart/x-mixed-replace; boundary=frame')

Sending every frame to clients can cause lots of unnecessary traffic transfer. It's suitable for live streaming, but may not suitable for monitor. What's more, it's easy to disconnect.

3. imageZMQ
advantage: real-time and distributable

disadvantage: you can't view frames on browser, you have to write/install a new program on your device.

4. ffmpeg + rtmp

5. websocket protocol
when server have frames to send to clients, it may:

• send frames url to clients using websocket protocol, then client(browser) execute img.src='http://domain.com/img/1.jpg'

  Modifying img.src too frequently would cause image url changed before it loaded, which lead to no image is shown. But, this is a good solution when the frame rate is very low.

• encode frames with base64, then send base64 string to clients using websocket protocol

  It cause meaningless calculation and increase amount of data to transfer.

• send frame binary(converted to jpg) to clients directly using websocket protocol. Then execute js shown below to display image.

   let blob = new Blob([msg], {type: 'image/jpeg'});
   img.src = window.URL.createObjectURL(blob);

This is the solution that I used.

Install

ubuntu

$ wget https://github.com/def-fun/rtm2/archive/master.zip
$ unzip master.zip
$ cd master/
$ pip3 install -r requirements.txt
$ gunicorn --worker-class eventlet -w 1 app:app -b 0.0.0.0:5000 

If terminal shows something like:

[ WARN:0] global /io/opencv/modules/videoio/src/cap_v4l.cpp (802) open VIDEOIO ERROR: V4L: can't open camera by index 0

it means you need to install right webCamera drive (note: you can get driver from http://www.ideasonboard.org/uvc/), or run command with sudo.

$ which gunicorn 
/home/def/.local/bin/gunicorn
$ sudo /home/def/.local/bin/gunicorn --worker-class eventlet -w 1 app:app -b 0.0.0.0:5000

windows

install python3 first, then download master.zip and uncompressed it. Open cmd

> cd master/
> pip install -r requirements.txt
> python app.py

Then visit http://127.0.0.1:5000/
(default username is user and password is change_it)

Detecting will start after the first visit. The default stream mode is detect mode, which can avoid lots of network traffic, as image shown below.

Todo

• optimize network transfer, for example, transfer changed area only.
• alarms or vibrate your device(mobile phone) when motion detected.
• beautify web page
• face recognition, YOLO, and so on

I'm have little experience in programing. So, current web page is ugly, many function haven't been finished, and English README need to be polished. I hope you can help me.

Pull requests are welcome :)

Similar_works

https://www.pyimagesearch.com/2019/04/15/live-video-streaming-over-network-with-opencv-and-imagezmq/ https://blog.miguelgrinberg.com/post/video-streaming-with-flask https://gist.github.com/n3wtron/4624820 https://github.com/Kr1s77/flask-video-streaming-recorder https://www.geeksforgeeks.org/webcam-motion-detector-python/ https://www.pyimagesearch.com/2015/06/01/home-surveillance-and-motion-detection-with-the-raspberry-pi-python-and-opencv/ https://www.pyimagesearch.com/2019/09/02/opencv-stream-video-to-web-browser-html-page/