DeviceWorker interface of DeepStream SDK in Jetson-TX2 (Solved)

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi,

I am just getting started with DeepStream SDK on Jetson-TX2. As a first step, I am trying to understand the sample application and get familiarized.

I came across this blog (https://devblogs.nvidia.com/deepstream-video-analytics-smart-cities/), and was able to get hold of it. The implementation uses DeviceWorker, Module and Tensor interface. I also could find the sample implementation in DeepStream SDK 1.5 for Tesla.

When I downloaded Deepstream SDK 1.5 for Jetson, I did’t find neither the 'deepstream.h’header file nor the same samples (as in the SDK of Tesla).Instead there was nvgstiva app which seems to use ‘gstreamer’ interface.

Questions:

  1. As a begginer, I found the DeviceWorker interface much intutive to understand. How can I run the sample appliation (say 'nvDecInfer_detection') on Jetson-TX2.
  2. Can you please guide me which is the right interface (gstreamer vs DeviceWorker) that I should be using if my platform is Jetson-TX2

Thanks in advance for the help. Looking forward for any responses.

2

Hi Vivek,
We don’t have NvDecInfer_detection support in Tx2, even on telsa, DeviceWorker interface has been depricated in recent release(Deepstream2.0).

On TX2, you can only use gstreamer interface.
BTW, on tx2, we have tegra_multimedia_api, you can also refer to that.

Thanks
wayne zhu

3

Hi WayneZhu,

Thanks for the clarification.

So, I ll use nvgstiva (gstreamer based app) as my reference. I can also check the ‘tegra_multimedia_api’ that you mentioned.