NVIDIA Research: Fast Uncertainty Quantification for Deep Object Pose Estimation

Originally published at: NVIDIA Research: Fast Uncertainty Quantification for Deep Object Pose Estimation | NVIDIA Developer Blog

Researchers from NVIDIA, University of Texas at Austin and Caltech developed a simple, efficient, and plug-and-play uncertainty quantification method for the 6-DoF object pose estimation task, using an ensemble of K pre-trained estimators with different architectures and/or training data sources.

Is it possible to use this on Jetson Platform ?
Could you show me its procedure for Jetson if possible ?

Our pose estimator models are based on DOPE (GitHub - NVlabs/Deep_Object_Pose: Deep Object Pose Estimation (DOPE) – ROS inference (CoRL 2018)). We have tested it on Ubuntu 20.04 with ROS Noetic with an NVIDIA Titan X and RTX 2080ti with Python 3.8. We expect this to work on the Jetson platform as well. You can check out the Issues on DOPE for pointers (e.g., Build Deep Object Pose on AGX · Issue #105 · NVlabs/Deep_Object_Pose · GitHub).