Nvidia releases FoundationPose on Hugging Face
HuggingPapers says the model estimates and tracks objects' 3D position and orientation and works on novel objects without fine-tuning.
TLDR
HuggingPapers says Nvidia has released FoundationPose on Hugging Face. It describes FoundationPose as a unified foundation model for six-degree-of-freedom (6-DoF) object pose estimation and tracking, claiming it works on novel objects without fine-tuning.
Combined views
594
1 Source, first seen ago