Keep it Unreal: Bridging the Realism Gap for 2.5D Recognition with Geometry Priors Only

@article{Zakharov2018KeepIU,
  title={Keep it Unreal: Bridging the Realism Gap for 2.5D Recognition with Geometry Priors Only},
  author={S. Zakharov and Benjamin Planche and Z. Wu and A. Hutter and H. Kosch and Slobodan Ilic},
  journal={2018 International Conference on 3D Vision (3DV)},
  year={2018},
  pages={1-11}
}
  • S. Zakharov, Benjamin Planche, +3 authors Slobodan Ilic
  • Published 2018
  • Computer Science
  • 2018 International Conference on 3D Vision (3DV)
  • With the increasing availability of large databases of 3D CAD models, methods for depth-based recognition of localized objects can be trained on an uncountable number of synthetically rendered images. [...] Key Method Purely trained on synthetic data, playing against an extensive augmentation pipeline in an unsupervised manner, our generative adversarial network learns to effectively segment depth images and recover the clean synthetic-looking depth information even from partial occlusions.Expand Abstract

    Figures, Tables, and Topics from this paper.

    Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimation
    35
    SyDD : Synthetic Depth Data Randomization for Object Detection using Domain-Relevant Background
    2
    Domain Transfer for 3D Pose Estimation from Color Images without Manual Annotations
    18
    CorNet: Generic 3D Corners for 6D Pose Estimation of New Objects without Retraining
    3
    DeceptionNet: Network-Driven Domain Randomization
    18
    CorNet: Generic 3D Corners for 6D Pose Estimation of New Objects without Retraining
    3

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 80 REFERENCES
    DepthSynth: Real-Time Realistic Synthetic Data Generation from CAD Models for 2.5D Recognition
    24
    Learning from Simulated and Unsupervised Images through Adversarial Training
    989
    Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Views
    481
    Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks
    690
    A deep representation for depth images from synthetic data
    26
    (DE)$^2$CO: Deep Depth Colorization
    17
    3D object instance recognition and pose estimation using triplet loss with dynamic margin
    14
    Improved Techniques for Training GANs
    3430