Security Fusion Node (SFN) is a platform to calibrate the camera stream and setup datastream flow to the various algorithms and direct the algorithm outputs to different ports.

The document outlines software requirements, camera calibration process and activation of the SFN to manage the datastreams. SFN typically receives data from VCA core. Please reffer to the VCA Installation tutorial for more details.

• Linux 16.04
• Nvidia drive 384 and above
• CUDA 9.0
• Docker
• MySQL

1. At the home directory, if not installed, then download WP5 Repo

git clone https://github.com/MONICA-Project/sfn.git

2. Install MySQL. Refer this link for detailed instructions on MySQL setup. Keep the username "root" and password "root".

sudo apt-get update
sudo apt-get install mysql-server
mysql_secure_installation (answer all question default, do not change the password)
mysqld --initialize (you will see an error, indicates file already exist)
mysqladmin -p -u root version (checks if MySQL is up and running)

3. Check if MySQL is up and runing and will be run automatically on boot.

systemctl status mysql.service
mysqladmin -p -u root version

4. If MySQL is not running, run followings, else ignore following:

sudo systemctl start mysql

5. Log in to the MySQL and check databases

mysql -u root -p
SHOW DATABASES;

6. Initialise WP5 mysql databases. (make sure no error pops)

mysql -u root -p < monica/WP5/KU/SecurityFusionNodeService/mysql_setup_db.sql

7. Log in to the mysql and check if 2 databases are installed (sfn_database & log_database)

sudo mysql -u root -p (password: root)
SHOW DATABASES; ( check the databases table)

To ensure all the required dependencies for SFN are available, install the following dependencies.

sudo apt install python3-pip
sudo pip3 install -r monica/WP5/KU/SecurityFusionNodeService/requirements.txt
sudo pip3 install -r monica/WP5/KU/KUConfigTool/requirements.txt
sudo pip3 install -r monica/WP5/KU/Algorithms/requirements.txt
sudo apt-get install python3-tk
sudo apt-get install vlc

If the SFN docker image is not built, then go to the monica folder and build the SFN image.

sudo docker build -f Dockerfile_sfn -t sfn .

The docker build takes a while and check the build messages for errors.

The docker image is run using sudo docker run --net="host" -t sfn. Keep the terminal open to see the transactions in the docker and any error messages.

The docker run command starts the SFN docker image. For the SFN to send algorithm observations to SCRAL, the calibrated camera needs to be registered by running the camera registration file.

Keep the SFN run terminal and in a separate terminal at cd monica location run the camera registration python file (xxx.py) using the command: python3 WP5/KU/SecurityFusionNodeService/xxx.py.

Refer to the camera calibration document for calibrating the camera and preparing the camera registration file.

The camera calibration process creates two files - a text file in KUConfigTool folder and .PK file in cam_config folder. Based on the .PK files, a camera registration file is to be created.

The registration file is a python file necessary to register the cameras to the SCRAL endpoint. It ensures the desired camera messages are forwarded.

The camera registration file is to be placed in the SecurityFusionNodeService folder. Use the template camera registration file already provided.

Two changes are particularly to be made/verified in the registration file. The SCRAL endpoint on the SFN. Ensure the correct camera config files, i.e. the .PK files are listed in the camera registration file.

The camera registration file is run using the following:

python3 WP5/KU/SecurityFusionNodeService/XXX_reg.py

Once the registration file is run, check the terminal where the SFN is running. If everything is correct, it shows the camera registration messages forwarded.

The camera registration file needs to be run everytime the SFN is restarted.

Contributions are welcome.

Please fork, make your changes, and submit a pull request. For major changes, please open an issue first and discuss it with the other authors.

This work is supported by the European Commission through the MONICA H2020 PROJECT under grant agreement No 732350.