DISP6D: Disentangled Implicit Shape and Pose Learning for Scalable 6D Pose Estimation

被引:6
|
作者
Wen, Yilin [1 ]
Li, Xiangyu [2 ]
Pan, Hao [3 ]
Yang, Lei [1 ,4 ]
Wang, Zheng [5 ]
Komura, Taku [1 ]
Wang, Wenping [6 ]
机构
[1] Univ Hong Kong, Hong Kong, Peoples R China
[2] Brown Univ, Providence, RI 02912 USA
[3] Microsoft Res Asia, Beijing, Peoples R China
[4] Ctr Garment Prod Ltd, Hong Kong, Peoples R China
[5] SUSTech, Shenzhen, Peoples R China
[6] Texas A&M Univ, College Stn, TX USA
来源
关键词
6D pose estimation; Scalability; Disentanglement; Symmetry ambiguity; Re-entanglement; Sim-to-real; REPRESENTATION;
D O I
10.1007/978-3-031-20077-9_24
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Scalable 6D pose estimation for rigid objects from RGB images aims at handling multiple objects and generalizing to novel objects. Building on a well-known auto-encoding framework to cope with object symmetry and the lack of labeled training data, we achieve scalability by disentangling the latent representation of auto-encoder into shape and pose sub-spaces. The latent shape space models the similarity of different objects through contrastive metric learning, and the latent pose code is compared with canonical rotations for rotation retrieval. Because different object symmetries induce inconsistent latent pose spaces, we re-entangle the shape representation with canonical rotations to generate shape-dependent pose codebooks for rotation retrieval. We show state-of-the-art performance on two benchmarks containing textureless CAD objects without category and daily objects with categories respectively, and further demonstrate improved scalability by extending to a more challenging setting of daily objects across categories.
引用
收藏
页码:404 / 421
页数:18
相关论文
共 50 条
  • [21] KDFNet: Learning Keypoint Distance Field for 6D Object Pose Estimation
    Liu, Xingyu
    Iwase, Shun
    Kitani, Kris M.
    2021 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2021, : 4631 - 4638
  • [22] NVR-Net: Normal Vector Guided Regression Network for Disentangled 6D Pose Estimation
    Feng, Guangkun
    Xu, Ting-Bing
    Liu, Fulin
    Liu, Mingkun
    Wei, Zhenzhong
    IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2024, 34 (02) : 1098 - 1113
  • [23] Learning 6D Object Pose Estimation Using 3D Object Coordinates
    Brachmann, Eric
    Krull, Alexander
    Michel, Frank
    Gumhold, Stefan
    Shotton, Jamie
    Rother, Carsten
    COMPUTER VISION - ECCV 2014, PT II, 2014, 8690 : 536 - 551
  • [24] SilhoNet: An RGB Method for 6D Object Pose Estimation
    Billings, Gideon
    Johnson-Roberson, Matthew
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2019, 4 (04): : 3727 - 3734
  • [25] Impact of Segmentation and Color Spaces in 6D Pose Estimation
    Pereira, Nuno
    Alexandre, Luis A.
    2021 IEEE INTERNATIONAL CONFERENCE ON AUTONOMOUS ROBOT SYSTEMS AND COMPETITIONS (ICARSC), 2021, : 228 - 233
  • [26] MagicCubePose, A more comprehensive 6D pose estimation network
    Fudong Li
    Dongyang Gao
    Qiang Huang
    Wei Li
    Yuequan Yang
    Scientific Reports, 13
  • [27] On Object Symmetries and 6D Pose Estimation from Images
    Pitteri, Giorgia
    Ramamonjisoa, Michael
    Ilic, Slobodan
    Lepetit, Vincent
    2019 INTERNATIONAL CONFERENCE ON 3D VISION (3DV 2019), 2019, : 614 - 622
  • [28] DeepIM: Deep Iterative Matching for 6D Pose Estimation
    Yi Li
    Gu Wang
    Xiangyang Ji
    Yu Xiang
    Dieter Fox
    International Journal of Computer Vision, 2020, 128 : 657 - 678
  • [29] EdgePose: An Edge Attention Network for 6D Pose Estimation
    Feng, Qi
    Nong, Jian
    Liang, Yanyan
    MATHEMATICS, 2024, 12 (17)
  • [30] Learning Analysis-by-Synthesis for 6D Pose Estimation in RGB-D Images
    Krull, Alexander
    Brachmann, Eric
    Michel, Frank
    Yang, Michael Ying
    Gumhold, Stefan
    Rother, Carsten
    2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, : 954 - 962