Lane follower algorithm for steering access control on Jetson TX2


Is there any lane follower algorithm for curved lanes using the concepts of lane-edge detection and slope calculation that is compatible with Jetson TX2 and works on real-time?
The algorithm needs to be in C++.

Kindly help.


Hi Pratosha, there is a simple lane detection demo included with VisionWorks samples, it uses Hough lines/curves and runs in realtime on TX1/TX2.

Hello Dusty,

Thanks for the information.
Is there any link available to access the source code?

Kindly help

The source is available on your Jetson after installing with JetPack. Or you can have JetPack download the packages to your host PC, then extract the VisionWorks sample .deb like a tar or zip archive, and the sources will be included inside of there.

Hello Dusty,

Yes we found the folder VisionWorks that contains the executable and source codes. We are facing issues while running the executable:

  1. Video not playing using Videos & VLC media player.
  2. unable to find /usr/share/visionworks/sources.

Is there any other algorithm available for lane detection in C++?


The lane detection demo source should be found at the following location:


If it isn’t there, are you sure that you had JetPack install the VisionWorks samples?

Hi Dusty,

Yes the Jetpack 3.1 was installed and VisionWorks was a part of it. The sfm executable is the only one present within the VisionWorks folder.
Is there any other way to install VisionWorks sources other than JetPack?



Checking with JetPack3.1, VisionWorks source code is located at ‘/usr/src/visionworks/sources/’ correctly.
Could you try to re-install VisionWorks package and try it again?


Hi AastaLLL,

Yes I would like to re- install VisionWorks.
Is there a link to install it without JetPack?
I dont want to flash the board with JetPack 3.1 again as few changes have been made.

Can we integrate Jetson-inference detectnet with lane detection using Hough transform visionworks on Jetson TX2?



Please use JetPack to install VisionWorks.
You can uninstall and re-install for VisionWorks only. No need to re-flashing the image.


  • Uninstall VisionWorks packages with the following command:
    $ sudo apt-get remove --purge libvisionworks-samples libvisionworks-dev libvisionworks-docs libvisionworks
  • Then uninstall local repositories:
    $ sudo apt-get purge libvisionworks-repo
  • Update apt cache after removing the repository:
    $ sudo apt-get update

Please use JetPack:

For integration:
We are not clear about the integration you want to do.
Could you describe more about the expected pipeline?


Is there any implementation in Python?