Corpus ID: 226964886

Fast Uncertainty Quantification for Deep Object Pose Estimation

  title={Fast Uncertainty Quantification for Deep Object Pose Estimation},
  author={G. Shi and Yifeng Zhu and J. Tremblay and Stan Birchfield and F. Ramos and A. Anandkumar and Yuke Zhu},
  • G. Shi, Yifeng Zhu, +4 authors Yuke Zhu
  • Published 2020
  • Computer Science
  • ArXiv
  • Deep learning-based object pose estimators are often unreliable and overconfident especially when the input image is outside the training domain, for instance, with sim2real transfer. Efficient and robust uncertainty quantification (UQ) in pose estimators is critically needed in many robotic tasks. In this work, we propose a simple, efficient, and plug-and-play UQ method for 6-DoF object pose estimation. We ensemble 2-3 pre-trained models with different neural network architectures and/or… CONTINUE READING

    Figures and Tables from this paper


    DPOD: 6D Pose Object Detector and Refiner
    • 61
    • PDF
    DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion
    • C. Wang, Danfei Xu, +4 authors S. Savarese
    • Computer Science
    • 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
    • 2019
    • 184
    • PDF
    Segmentation-Driven 6D Object Pose Estimation
    • 61
    • PDF
    Multi-view self-supervised deep learning for 6D pose estimation in the Amazon Picking Challenge
    • 237
    • PDF
    Deep Object Pose Estimation for Semantic Robotic Grasping of Household Objects
    • 180
    • PDF
    PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenes
    • 493
    • PDF
    A General Framework for Uncertainty Estimation in Deep Learning
    • 30
    • PDF
    Sim2real transfer learning for 3D human pose estimation: motion to the rescue
    • 22
    • PDF
    Real-Time Seamless Single Shot 6D Object Pose Prediction
    • 285
    • PDF