DeepStream 5.1 Human Pose Estimation & Cuda 11

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I try to run Human Pose Estimation on Deepstream 5.1 from GitHub - NVIDIA-AI-IOT/deepstream_pose_estimation: This is a sample DeepStream application to demonstrate a human pose estimation pipeline.
My problem that I have RTX 3060 and only CUDA 11.* support that device, but Pose Estimation work with only CUDA 10.2. When I use CUDA 11 - model can not start because can not find file ****.10.2 - I assume that file only in CUDA 10.2 package.
When I use CUDA 10.2 with Pytorch - “Graphics Device with CUDA capability sm_86 is not compatible with the current PyTorch installation.
The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_70”
How I can find solution for that problem?

3

Hey customer,

  1. please share your setup with us for better support
  2. I think you need to replace your libnvds_osd.so with libnvds_osd_cuda11.so