Metron AI - Artificial Data Generator (ArDaGen)

Metron AI ArDaGen is a synthetic data generator build on NVIDIA Omniverse platform to generate physically accurate images and annotations to train Metron AI.

Explore Metron AI ArDaGen docs »

Key Features

💥 photorealistic physical based rendering via NVIDIA Omniverse 💥 vast image content variations 💥 100% accurate annotations 💥 uses SOTA configuration framework - Meta's Hydra 💥 modular architecture with custom design scenarios

Table of Contents

• How to run
• Status
• Documentation
  • Building & running documentation locally

How to run

After you install the solution, go into Isaac Sim root folder in a terminal and run

./python.sh metron_ai_ardagen/metron_ai_ardagen_run.py

It loads ArDaGen's configuration via Hydra framework. See configuration section for mode details.

Status

Documentation

Online documentation of the latest commit on the main branch can be found here. If you don't want to build a new documentation and use the build in the repository, go to step 5 directly.

Building & running documentation locally

1. Download the repository using git clone https://github.com/OndrejSzekely/metron_ai_ardagen.git

2. It is recommended to use virtual environment, to encapsulate the dev tools (Python frameworks and other software components). Create Python 3.7 virtual environment using Python dependency management tool you are using (e.g. Conda, Pipenv, etc...).

   💡 Reference Installation Steps 👇

   It is recommended to use Anaconda channel (how to get Anaconda), which also provides installation management of non-Python software components, and Python. Run following command to create a new virtual environment:

   conda create -n metron_dev_37 python=3.7

   Run following command to attach created virtual environment in which all further steps are executed:

   conda activate metron_dev_37

3. Install following frameworks in the environment.

   sphinx (~5.0.2)
   myst-parser (=1.0.0)
   sphinxemoji (=0.2.0)
   sphinx-design (=0.3.0)
   sphinx-book-theme (=1.0.0)
   sphinx-copybutton (=0.5.1)
   

   💡 Reference Installation Steps 👇

   In the activated environment run following commands:

   conda install -y -c anaconda sphinx=5.0.2
   conda install -y -c conda-forge myst-parser=1.0.0 sphinx-copybutton=0.5.1 sphinxcontrib-mermaid=0.8.1 sphinx-subfigure=0.2.4
   pip install -r requirements_docs.txt

4. Go into repository's root folder and in the activated environment build the documentation:

   sphinx-build -b html docs_src docs

   💡 Reference Installation Steps 👇

   In the activated environment run following command:

   sphinx-build -b html docs_src docs

5. In repository's root folder run a HTTP server with the documentation:

   python -m http.server --directory docs 4444

   Then open your browser http://localhost:4444 and see the documentation.

   GOOD JOB! 🙌 🚀 💫

   💡 Reference Installation Steps 👇

   In repository's root folder and activated environment run a HTTP server with the documentation:

   python -m http.server --directory docs 4444

   Then open your browser http://localhost:4444 and see the documentation.