Resolving Symmetry Ambiguity in Correspondence-Based Methods for Instance-Level Object Pose Estimation

被引:0
作者
Lin, Yongliang [1 ,2 ]
Su, Yongzhi [3 ,4 ]
Inuganti, Sandeep
Di, Yan [5 ]
Ajilforoushan, Naeem [1 ,2 ]
Yang, Hanqing [1 ,2 ]
Zhang, Yu [1 ,2 ]
Rambach, Jason [6 ]
机构
[1] Zhejiang Univ, State Key Lab Ind Control Technol, Hangzhou 310027, Peoples R China
[2] Zhejiang Univ, Coll Control Sci & Engn, Hangzhou 310027, Peoples R China
[3] German Res Ctr Artificial Intelligence DFKI GmbH, German Res Cen-ter Artificial Intelligence (DFKI G, D-67663 Kaiserslautern, Germany
[4] RPTU Kaiserslautern, Fac Comp Sci, D-67663 Kaiserslautern, Germany
[5] Tech Univ Munich, Fac Comp Sci, D-85748 Garching, Germany
[6] DFKI GmbH, D-67663 Kaiserslautern, Germany
关键词
Three-dimensional displays; Image coding; Pose estimation; Solid modeling; Training; Image color analysis; Encoding; Binary codes; Accuracy; Translation; Object pose estimation; symmetry ambiguity; deep learning for visual perception; representation learning;
D O I
10.1109/TIP.2025.3544142
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Estimating the 6D pose of an object from a single RGB image is a critical task that becomes additionally challenging when dealing with symmetric objects. Recent approaches typically establish one-to-one correspondences between image pixels and 3D object surface vertices. However, the utilization of one-to-one correspondences introduces ambiguity for symmetric objects. To address this, we propose SymCode, a symmetry-aware surface encoding that encodes the object surface vertices based on one-to-many correspondences, eliminating the problem of one-to-one correspondence ambiguity. We also introduce SymNet, a fast end-to-end network that directly regresses the 6D pose parameters without solving a PnP problem. We demonstrate faster runtime and comparable accuracy achieved by our method on the T-LESS and IC-BIN benchmarks of mostly symmetric objects. The code is available at https://github.com/lyltc1/SymNet.
引用
收藏
页码:1700 / 1711
页数:12
相关论文
共 45 条
[1]   Trends and challenges in robot manipulation [J].
Billard, Aude ;
Kragic, Danica .
SCIENCE, 2019, 364 (6446) :1149-+
[2]   SilhoNet: An RGB Method for 6D Object Pose Estimation [J].
Billings, Gideon ;
Johnson-Roberson, Matthew .
IEEE ROBOTICS AND AUTOMATION LETTERS, 2019, 4 (04) :3727-3734
[3]  
Bukschat Y, 2020, Arxiv, DOI [arXiv:2011.04307, DOI 10.48550/ARXIV.2011.04307]
[4]   CRT-6D: Fast 6D Object Pose Estimation with Cascaded Refinement Transformers [J].
Castro, Pedro ;
Kim, Tae-Kyun .
2023 IEEE/CVF WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV), 2023, :5735-5744
[5]  
Chen LC, 2017, Arxiv, DOI arXiv:1706.05587
[6]   SO-Pose: Exploiting Self-Occlusion for Direct 6D Pose Estimation [J].
Di, Yan ;
Manhardt, Fabian ;
Wang, Gu ;
Ji, Xiangyang ;
Navab, Nassir ;
Tombari, Federico .
2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, :12376-12385
[7]  
Do T.-T., 2019, P BRIT MACH VIS C, P1
[8]  
Do TT, 2018, Arxiv, DOI [arXiv:1802.10367, 10.48550/ARXIV.1802.10367]
[9]   Recovering 6D Object Pose and Predicting Next-Best-View in the Crowd [J].
Doumanoglou, Andreas ;
Kouskouridas, Rigas ;
Malassiotis, Sotiris ;
Kim, Tae-Kyun .
2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2016, :3583-3592
[10]   Efficient Center Voting for Object Detection and 6D Pose Estimation in 3D Point Cloud [J].
Guo, Jianwei ;
Xing, Xuejun ;
Quan, Weize ;
Yan, Dong-Ming ;
Gu, Qingyi ;
Liu, Yang ;
Zhang, Xiaopeng .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2021, 30 :5072-5084