Difference between the offline and non-offline versions of the PhysicalAI-Autonomous-Vehicles dataset

#41
by ethentsao-pronto - opened

Hi NVIDIA team,

I'm using the PhysicalAI-Autonomous-Vehicles dataset and have a question about the offline and non-offline versions mentioned in the version history.

Could you clarify:

  • What processing pipeline is applied to generate the offline/non-offline version?
  • Does the offline version use a different calibration method?
  • Which version would you recommend for training perception models?

I'd appreciate any details or references to documentation describing the differences between the two versions.

Thanks in advance!

Sign up or log in to comment