UniPose9D: Universal Category-Agnostic Object Pose Estimation

Published in arXiv, 2026

Yang You, Yi Du, Cole Harrison, Leonidas Guibas

arXiv / PDF / code / project page / BibTeX

Abstract

UniPose9D is a category agnostic foundation model for 9D object pose estimation. Given an object mask and either a color and depth observation or a color image with predicted depth, it estimates rotation, translation, and metric size without category labels, CAD models, shape priors, or reference views. A single model performs competitively across standard benchmarks and transfers to unseen objects and everyday scenes.

BibTeX

@article{you2026unipose9d,
  title     = {UniPose9D: Universal Category-Agnostic Object Pose Estimation},
  author    = {Yang You and Yi Du and Cole Harrison and Leonidas Guibas},
  journal   = {arXiv preprint arXiv:2607.09985},
  year      = {2026},
  eprint    = {2607.09985},
  archivePrefix = {arXiv}
}

Recommended citation: You, Y., Du, Y., Harrison, C., & Guibas, L. (2026). UniPose9D: Universal Category-Agnostic Object Pose Estimation. arXiv preprint arXiv:2607.09985.