This repo includes a simple program that shows how to deliver a video stream using Kombu and RabbitMQ in real-time, with an assumption that the sender and the receiver reside in the same computer system.

RabbitMQ is an open-source message broker.

Kombu is a python-based messaging library that can use RabbitMQ as a transport.

1. Install RabbitMQ and run the server.

   • Installation guides for difference OS systems including Linux, Windows and Mac OS X are available at https://www.rabbitmq.com/download.html. On Mac OS X, you can install RabbitMQ and run the servcer as a service as follows.

     brew install rabbitmq
     brew services start rabbitmq
     

   • After starting the server, you can access the server monitoring panel at http://localhost:15672

2. Install Kombu

   • Installation guide is available at https://kombu.readthedocs.io/en/latest/introduction.html#installation. In most systems, it is as simple as:

     pip install kombu
     

3. Install OpenCV and Python Wrapper.

   • Refer to the guide at https://pypi.org/project/opencv-python/. In most systems, it is simply installed by typing:

     pip install opencv-python
     

1. Clone this repository and cd into the repository folder.

   git clone https://github.com/zebehn/kombu-video-streaming.git
   cd kombu-video-streaming
   

2. Run the video stream consumer. As there is no message published yet, it just hangs around.

   python video_consumer.py
   

3. Run the video stream producer. You will see the webcam lights on, a window pops up and shows a video stream captured by your webcam if successful.

   python video_producer.py
   

• Set the delivery_mode of Kombu's Exchange object to 1, which sets it to a transient mode that does not write message data to disk.
• Shrink image size. With my system, the original webcam image size is 1280x720, and when I resized it to 768x432, I got real-time delivery (29fps publish, 29 delivery, 29 consumer acks, 0 unacked, 0 disk writes)

If you have any suggestions, bug reports or problems, please report them to the issue tracker.

You are highly encouraged to improve, refactor or add your idea to this program. Just make pull requests!

This software is licensed under the New BSD-3 License. See the LICENSE file in the top distribution directory for the full license text.