How to preprocess image and postprocess results from BodyPoseNet?

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

Hi,

I come across BodyPoseNet on BodyPoseNet | NVIDIA NGC, but i find the information on model card quite lacking. I have a few questions as below:

  1. What is the recommended input image size for this model?
  2. Is there an existing preprocessing and postprocessing python script to refer from NVIDIA-AI-IOT/tao-toolkit-triton-apps: Sample app code for deploying TAO Toolkit trained models to Triton (github.com), else can you point me to the right resources?

Thank you

3

You can refer to tao_tensorflow1_backend/nvidia_tao_tf1/cv/bpnet/scripts at main · NVIDIA/tao_tensorflow1_backend · GitHub.