Is there getting started for DeepStream SDK 4.0?

Dear all,

I need step by step from the beginning how to use DeepStream SDK 4.0, such as hello world, accessing a camera, etc.
Thank you very much in advance.

warmest regards,
suryadi

Hi suryadi,

There is no tutorial ready for beginner yet, but you can find some information from https://developer.nvidia.com/deepstream-sdk

Thanks

Hi suryadi,

The best way to get started on DeepStream is go thru the DeepStream development guide which will walk you thru how to run the sample app. https://docs.nvidia.com/metropolis/deepstream/4.0/dev-guide/

Once you do that go to ‘sources/apps/sample_apps/deepstream-test1’. This is the basic ‘hello world’ for DeepStream. Take a look at the README and go thru the code. It provides reference to build basic the pipeline with DeepStream plugins.
For more info on examples: DeepStream reference examples

After going thru test1, take a look at test2, test3 and test4. All the subsequent tests builds on test1 with different functionality. test4 will walk you to connecting to cloud.

If you have any questions, please post your questions on this forum.

Thanks,
Chintan

hi Chintan,

I posted this question in a different way in the sdk forum, please let me know

I want to attach camera to tesla v100,

  1. can we export a trained model from tenforflow using keras to the tensorRT inference engine?
  2. Looks like ngc container is based out of C? Can we only use models provided by the library?
  3. what is the purpose of test1 to test5 samples ? can you explain how to test inferencing here?
  4. can we do inferecing the jetson way?
  5. since I downloaded the ngc container for tesla, does it not come with all the prebuilt libraries like
    gstreamer etc? Looks like we have to download these libraries…

Thanks,
gayatri