Defender

Note: This project is no longer actively maintained. No support is provided for bugfixes, security updates, etc. It was originally started as a hobby, and built around Python versions 3.4-3.6, and original Angular. It has many examples for processing images, video, audio, HTTP requests, web apps, etc., although likely out of date. The project also contains a good baseline for code quality standards on new projects, so it remains available for referencing.

Defender is a Python and Javascript based Web Application/Rest API to manage streaming of audio and video from security cameras, and other devices, in a local network.

Security is currently achieved through a combination of HTTPS certificates, and a salted User/Password database with PBKDF2 SHA512 combo.

For simplicity and greater compatibility with low-end devices, video is streamed as MJPEG, and audio is streamed as WAV (meaning bandwidth usage is not optimized). Streams can be saved using separate applications with support for HTTP, such as FFMPEG, VLC, etc.

This application supports Video streaming through OpenCV:
http://opencv.org/

And audio through PyAudio:
https://pypi.python.org/pypi/PyAudio

Requirements

• Python3.6+
• System with video and/or audio capture device

Recommendations

• Raspberry Pi 3 or equivalent hardware
• Webcam with builtin microphone. Tested on Logitech C210.

Quick Start

1. Setup OpenCV and Python environment with:
   How to set up Raspberry Pi

2. Copy frontend demo data to local folder:

   cp -R defender/demodata ~/.defender
   cp -R defender/html ~/.defender
   

3. Update the configuration file with values to match your environment and devices' capabilities:

   defender/scripts/defend.py --list-devices
   v4l2-ctl --list-devices
   vi ~/.defender/config.json
   

4. Generate new custom HTTPS certificate:

   openssl req -newkey rsa:4096 -nodes -keyout ~/.defender/key.pem -x509 -days 365 -out ~/.defender/server.pem
   

5. If setup for 'virtualenv' per Step 1, ensure you are working on that project:

   workon opencv
   

6. Start server with:

   defender/scripts/defend.py -c ~/.defender/config.json
   

7. Add a user and password to the database:

   # Skip add to use default:default
   user add <new user name>
   
   # If new user was added, remove default:
   user remove default
   

8. Attempt to load the video stream in a browser by going to:

   https://<hostname or ip>:<port>/video
   

Known Limitations

• For some reason HTTP traffic appears to fail loading page/serving API calls, until reason is determined HTTPS should always be used. Well, HTTPS should always be used.
• Multiple concurrent connections to audio streams may overflow buffer and degrade playback.
• Password hashing can lead to slowdown with the initial load of the provided HTML frontend unless the JS scripts are minified into one.

What does the future hold?

Note: This project is no longer actively maintained. No support is provided for bugfixes, security updates, etc. It was originally started as a hobby, and built around Python versions 3.4-3.6, and original Angular. It has many examples for processing images, video, audio, HTTP requests, web apps, etc., although likely out of date. The project also contains a good baseline for code quality standards on new projects, so it remains available for referencing.

• Multiple streams per device.
• Motion tracking.
• Audio Playback (two way communication).
• Support for controlling remote motorized devices, such as nerf turret cameras, drones, etc.
• Builtin A/V recording through application, instead of with 3rd party applications (FFMPEG/VLC).
• Possibly additional compression formats for A/V streams.
• Migrate HTTP interfaces to Flask/FastAPI.
• Implement API sessions to increase authorization speed on lower end devices.