CASIA OpenIR  > 智能机器人系统研究
6D Object Pose Estimation using Few-Shot Instance Segmentation and 3D Matching
Wanyi Li; Jia Sun; Yongkang Luo; Peng Wang
2019-09-25
Conference Name2019 IEEE Symposium Series on Computational Intelligence (SSCI)
Conference Date2019.12.6-9
Conference PlaceXiamen, China
Abstract

6D object pose estimation is an important but difficult computer vision task. It has many applications such as robotic manipulation and augmented reality. Although a large number of 6D object pose estimation methods have been developed, there are still many challenges, for example, background clutter, foreground occlusion, and lack of annotated training samples. To deal with these difficulties, a compact and effective algorithm for 6D pose estimation using RGB-D data under few-shot condition is presented in this paper. The proposed algorithm consists of two stages. The first stage is fewshot instance segmentation, which segments known objects from RGB image. The second stage is 3D matching, which recovers the poses of objects from cropped point clouds. Proposed segmentation method can achieve satisfactory performance using only few labeled samples. Comparison experiments on two challenging datasets are carried out, and the results demonstrate that the proposed method outperforms the state-of-the-art greatly. Recall scores obtained by the proposed method are 74.91% and 55.44%, while of the state-of-the-art are 61.87% and 44.92%, obtaining 13.04% and 10.52% improvement respectively.

Indexed ByEI
Funding ProjectNational Natural Science Foundation of China[91748131] ; National Natural Science Foundation of China[U1613213] ; National Natural Science Foundation of China[61771471] ; Youth Innovation Promotion Association of CAS[2015112]
Language英语
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/25850
Collection智能机器人系统研究
Corresponding AuthorYongkang Luo
AffiliationInstitute of Automation, Chinese Academy of Sciences, Beijing 100190, China
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Corresponding Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Wanyi Li,Jia Sun,Yongkang Luo,et al. 6D Object Pose Estimation using Few-Shot Instance Segmentation and 3D Matching[C],2019.
Files in This Item: Download All
File Name/Size DocType Version Access License
IEEE SSCI 2019-CIMSI(1436KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Wanyi Li]'s Articles
[Jia Sun]'s Articles
[Yongkang Luo]'s Articles
Baidu academic
Similar articles in Baidu academic
[Wanyi Li]'s Articles
[Jia Sun]'s Articles
[Yongkang Luo]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Wanyi Li]'s Articles
[Jia Sun]'s Articles
[Yongkang Luo]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: IEEE SSCI 2019-CIMSIVP_ieeeconf_letter_190925.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.