NVIDIA Omniverse Replicator Generates Synthetic Training Data for Robots

Originally published at: NVIDIA Omniverse Replicator Generates Synthetic Training Data for Robots | NVIDIA Developer Blog

Synthetic Data generation cockpit for machine learning (ML) engineers.

@jwitsoe
I have done quit a few tutorials on TAO (TLT3) and I would like to do one that uses images produced by NVIDIA Omniverse Replicator for Isaac Sim. I like the Forklift example.
Can you point me in the direction of resources on how to use the NVIDIA Omniverse Replicator for Isaac Sim

The upcoming release of Isaac Sim will contain all of the new synthetic data features used in the forklift example. Stay tuned for that release which is scheduled to happen soon.

I attempted to install Omniverse and the Issac sim but since I only have a GTX 1080TI it doesn’t work.
Do you have any recommendations for a cloud computing service has the RTX 3070 capability’s

Have you seen these docs?

[Native Workstation Deployment — Omniverse Robotics documentation]

https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/setup.html#cloud-deployment

Exactly what I was looking for

Thanks

Is the synthetic data generated by ISSAC Synthetic data generator annotated and in the KITTI format?

Will omniverse run on a RTX 4000

The minimum specs say RTX 3070. Does that include the Quadro RTX series

Yes, Isaac Sim can generate data in the KITTI format. Read More: Synthetic Data — Omniverse Robotics documentation

Now that Issac Sim 2021.2.0 is out.
Yeah
Where can I find the tutorial associated to this Developer Blog post?
" NVIDIA Omniverse Replicator Generates Synthetic Training Data for Robots"

Replicator Composer was used to generate the datasets referred to in the blog post. Documentation on this tool can be found here. 7. Replicator Composer — Omniverse Robotics documentation