NVIDIA-ISAAC-ROS / NVIDIA-ISAAC-ROS/isaac_ros_pose_estimation

centerpose_converter.py is missing for converting pth to onnx inside ros2 docker

Open
#67 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
501
Forks
57
PR merge metrics
No merged PRs in 30d

Description

Similiar to dope_converter.py, could you please share the script for centerpose_converter.py for converting the pth model trained using TAO toolkit to ONNX model inside the ros2 docker?

From isaac_ros_dope tutorial:

Image

ros2 run isaac_ros_dope dope_converter.py --format onnx \ --input ${ISAAC_ROS_WS}/isaac_ros_assets/models/dope/Ketchup.pth --output ${ISAAC_ROS_WS}/isaac_ros_assets/models/dope/Ketchup.onnx --row 720 --col 1280

https://nvidia-isaac-ros.github.io/repositories_and_packages/isaac_ros_pose_estimation/isaac_ros_dope/index.html#quickstart

Currently the tutorial for isaac_ros_centerpose directly downloads the onnx from ngc website. I need to be able to reproduce the results for my custom object class.

Probably same or slight modification of https://github.com/NVIDIA/tao_pytorch_backend/blob/main/nvidia_tao_pytorch/cv/centerpose/scripts/export.py

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by comparing the existing dope_converter.py entry point with NVIDIA's TAO export.py linked in the issue, then review the isaac_ros_centerpose tutorial and its ROS 2 Docker context. Done means a centerpose_converter.py workflow can convert a TAO-trained .pth model to ONNX inside the container for a custom object class.

Written by the indexing model from the issue text.

Assessment

Tech stack
docker, python
Domain
devops, machine-learning, robotics
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.