NextFin

Noitom Robotics Opens HiPHI-Hand Dataset for NVIDIA EgoHand

Summarized by NextFin AI
  • Noitom Robotics, a physical AI data company, publicly released its high-precision hand motion-capture dataset HiPHI-Hand on Hugging Face.
  • The dataset was used to train NVIDIA EgoHand, a monocular 3D hand reconstruction model built on the NVIDIA Isaac Capture framework.
  • EgoHand can accurately reconstruct hand pose and shape from first-person video captured by a single camera.
  • The release provides synchronized egocentric video and sub-millimeter optical motion-capture ground truth for egocentric vision, hand reconstruction and robot learning research.

NextFin News — Physical AI data company Noitom Robotics has publicly released its high-precision hand motion-capture dataset HiPHI-Hand on Hugging Face.

The dataset was used to train NVIDIA EgoHand, a monocular 3D hand reconstruction model built on the NVIDIA Isaac Capture framework. EgoHand can accurately reconstruct hand pose and shape from first-person video captured by a single camera.

The release provides synchronized egocentric video and sub-millimeter optical motion-capture ground truth for research in egocentric vision, hand reconstruction and robot learning.

Explore more exclusive insights at nextfin.ai.

Search, ask or analyze anything...