Hand PointNet : 3D hand pose estimation using point sets
Convolutional Neural Network (CNN) has shown promising results for 3D hand pose estimation in depth images. Different from existing CNN-based hand pose estimation methods that take either 2D images or 3D volumes as the input, our proposed Hand PointNet directly processes the 3D point cloud that mode...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/88581 http://hdl.handle.net/10220/45084 http://openaccess.thecvf.com/content_cvpr_2018/html/Ge_Hand_PointNet_3D_CVPR_2018_paper.html |
| _version_ | 1826179027838173184 |
|---|---|
| author | Ge, Liuhao Cai, Yujun Weng, Junwu Yuan, Junsong |
| author2 | Interdisciplinary Graduate School (IGS) |
| author_facet | Interdisciplinary Graduate School (IGS) Ge, Liuhao Cai, Yujun Weng, Junwu Yuan, Junsong |
| author_sort | Ge, Liuhao |
| collection | NTU |
| description | Convolutional Neural Network (CNN) has shown promising results for 3D hand pose estimation in depth images. Different from existing CNN-based hand pose estimation methods that take either 2D images or 3D volumes as the input, our proposed Hand PointNet directly processes the 3D point cloud that models the visible surface of the hand for pose regression. Taking the normalized point cloud as the input, our proposed hand pose regression network is able to capture complex hand structures and accurately regress a low dimensional representation of the 3D hand pose. In order to further improve the accuracy of fingertips, we design a fingertip refinement network that directly takes the neighboring points of the estimated fingertip location as input to refine the fingertip location. Experiments on three challenging hand pose datasets show that our proposed method outperforms state-of-the-art methods. |
| first_indexed | 2024-10-01T03:30:44Z |
| format | Conference Paper |
| id | ntu-10356/88581 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T03:30:44Z |
| publishDate | 2018 |
| record_format | dspace |
| spelling | ntu-10356/885812020-11-01T04:43:13Z Hand PointNet : 3D hand pose estimation using point sets Ge, Liuhao Cai, Yujun Weng, Junwu Yuan, Junsong Interdisciplinary Graduate School (IGS) School of Electrical and Electronic Engineering The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2018 Institute for Media Innovation 3D Hand Pose Pose Regression Convolutional Neural Network (CNN) has shown promising results for 3D hand pose estimation in depth images. Different from existing CNN-based hand pose estimation methods that take either 2D images or 3D volumes as the input, our proposed Hand PointNet directly processes the 3D point cloud that models the visible surface of the hand for pose regression. Taking the normalized point cloud as the input, our proposed hand pose regression network is able to capture complex hand structures and accurately regress a low dimensional representation of the 3D hand pose. In order to further improve the accuracy of fingertips, we design a fingertip refinement network that directly takes the neighboring points of the estimated fingertip location as input to refine the fingertip location. Experiments on three challenging hand pose datasets show that our proposed method outperforms state-of-the-art methods. NRF (Natl Research Foundation, S’pore) MOE (Min. of Education, S’pore) Accepted version 2018-07-16T06:19:57Z 2019-12-06T17:06:33Z 2018-07-16T06:19:57Z 2019-12-06T17:06:33Z 2018 Conference Paper https://hdl.handle.net/10356/88581 http://hdl.handle.net/10220/45084 http://openaccess.thecvf.com/content_cvpr_2018/html/Ge_Hand_PointNet_3D_CVPR_2018_paper.html en © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The published version is available at: [http://openaccess.thecvf.com/content_cvpr_2018/html/Ge_Hand_PointNet_3D_CVPR_2018_paper.html]. 10 p. application/pdf |
| spellingShingle | 3D Hand Pose Pose Regression Ge, Liuhao Cai, Yujun Weng, Junwu Yuan, Junsong Hand PointNet : 3D hand pose estimation using point sets |
| title | Hand PointNet : 3D hand pose estimation using point sets |
| title_full | Hand PointNet : 3D hand pose estimation using point sets |
| title_fullStr | Hand PointNet : 3D hand pose estimation using point sets |
| title_full_unstemmed | Hand PointNet : 3D hand pose estimation using point sets |
| title_short | Hand PointNet : 3D hand pose estimation using point sets |
| title_sort | hand pointnet 3d hand pose estimation using point sets |
| topic | 3D Hand Pose Pose Regression |
| url | https://hdl.handle.net/10356/88581 http://hdl.handle.net/10220/45084 http://openaccess.thecvf.com/content_cvpr_2018/html/Ge_Hand_PointNet_3D_CVPR_2018_paper.html |
| work_keys_str_mv | AT geliuhao handpointnet3dhandposeestimationusingpointsets AT caiyujun handpointnet3dhandposeestimationusingpointsets AT wengjunwu handpointnet3dhandposeestimationusingpointsets AT yuanjunsong handpointnet3dhandposeestimationusingpointsets |