Traffic Intelligence Tools

Turn key solution for compilation and deployment of Traffic Intelligence Tools with opencv 3.4.3 on Windows platform.

Install required programs

After cloning the project, you need to run the install script as Administrator.

Following programs will be installed onto your windows system:

• git -- for code download
• vagrant -- virtualbox VM manager
• virtualbox -- ubuntu 16.04 virtual machine
• mobaxterm -- running X-app on windows

Compiling opencv 3.4.3 and Traffic Intelligence

After required app installed, the script will create a directory cvtivm at the parent folder and copy bootstrap files to start a Ubuntu 16.04 VM as a host to compile opencv and traffic intelligence software.

The compilation usually will take 3-5 hours. After compilation is done opencv 3.4.3 and traffic intelligence debian installation packages will be created and copied to the release folder. Those packages can be installed on ubuntu 16.04 platform.

Using traffic intelligence on windows

Inside the cvtivm folder three directories are created:

• release -- compiled ubuntu packages copied here
• test -- traffic intelligence test package
• jupyter -- jupyter notebook sample for traffic intelligence test and jupyter folders are mirrored (shared) in the virtual machine.

Run display-trajectories.py demo

The display-trajectories.py demo has GUI element. We need to use mobaxterm to start a X server then login into the VM using ssh-config. If in the directory, vagrant-ssh is not there, you need to generate with following command:

• generate vagrant-ssh

vagrant ssh-config>vagrant-ssh

• vagrant-ssh sample:

Host cvti
  HostName 127.0.0.1
  User vagrant
  Port 2222
  UserKnownHostsFile /dev/null
  StrictHostKeyChecking no
  PasswordAuthentication no
  IdentityFile /cygdrive/c/dev/TrafficIntelligenceTool/cvtivm/.vagrant/machines/cvti/virtualbox/private_key
  IdentitiesOnly yes
  LogLevel FATAL
  ForwardAgent yes
  ForwardX11 yes

• Launch mobaxterm launch a local term (x server will start automatically on windows) mobaxterm
• find the directory where you have your virutal machine running, in my case /drive/c/dev/TrafficIntelligenceTool/cvtivm
• ssh into the VM with X-forward usig following command

ssh -F vagrant-ssh cvti

• here cvti is the host name specified in the vagrant-ssh configuration file
• unzip the demo at ~/test/

unzip 12-laurier.zip
cd 12-laurier

• Launch the demo with following command:

display-trajectories.py --cfg tracking.cfg -i laurier.avi -d laurier.sqlite -o homography.txt -t object

• A video screen will pop up with the first frame, mouse select the frame and hit return and each return key hit will load a frame from the video and draw trajectory information on the video frame.

Jupyter notebook sample

A jupyter notebook is at the jupyter shared directory. Launch your browser and access the sample at

http://cvti.example.com:8888

Traffic Intelligence Tool documation