Input data type on deepstream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

• Hardware Platform (Jetson / GPU) : Jetson nano
• DeepStream Version : 6.0
• JetPack Version (valid for Jetson only)
• TensorRT Version : 8.0.1
• NVIDIA GPU Driver Version (valid for GPU only) 10.2
• Issue Type( questions, new requirements, bugs) : Question

GitHub - NVIDIA-AI-IOT/deepstream_pose_estimation: This is a sample DeepStream application to demonstrate a human pose estimation pipeline..
Refer to above documentation, i want to customize deepstream pipeline to predict human Gesture from above pipeline.
the above pipeline of PGIE output is 18 points of pose dot values, and what i want to run in SGIE model is stacking the latest pose dot values and then run the SGIE model to predict gesture values.
so, i’m ready to run SGIE model but i dont have any idea how to input data like dot values or vector values. [also stacking latest dot points is needed too…]
anybody have idea about my question?? pls help…

3

Is it possible to put those data into user metadata?

4

i didn’t look for how to put those data into user metadata, but
if i put those stacking dot points to user metadata, then is it possible to run tensorrt Engine with user metadata on deepstream pipeline???

5
  • This question will be similar to my above question, then @AastaLLL , i think you can answer to my questions… thanks
6

hello… you guys have no solutions about my question???

7

Infer plugin is open source plugin. You can change it based on your requirement.