Unable to convert Image types


Does anyone know how I can convert vx_image to nvxcu_image_t so that even the contents are transferred. The structure of nvxcu_image_t doesn’t seem to have a parameter for context.

Any information on this will be of great help! Thanks in advance!


In visionworks sample code located at /usr/share/visionworks/sources
There are two sample to demonstrate feature tracking, one use vx_image and the other use nvxcu_image_t.

In cuda layer API (nvx_demo_feature_tracker_nvxcu), it also shows how to create frame source and render.
Maybe this sample code can help your problem.


Thank you for your reply. I am trying to implement remap function nvxcuRemap and one of the parameters is the remap table which is in the image format (nvxcu_image_t). I can populate the remap table without cuda and get it in vx_remap format but how do I convert this to nvxcu_image_t andd populate it with the values calculated? None of the examples shows how to get that done.



nvxcu_image_t is cuda API.

In Visionworks document, there is a tutorial for NVX CUDA API.

It can guide you how to access cuda data (ex. nvxcu_pitch_linear_image_t…) for your use-case.

After reading the VisionWorks’ documentation I haven’t figure it out how to convert from vx_image to nvxcu_image_t, or from a openCV image to a nvxcu_image_t image.
Also, is there a way to capture udp streaming using VisionWorks?

Hi agallegos,

Please file a new topic for your issue.