PointPoseNet: Point Pose Network for Robust 6D Object Pose Estimation

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Standard

PointPoseNet : Point Pose Network for Robust 6D Object Pose Estimation. / Chen, Wei; Duan, Jinming; Basevi, Hector; Chang, Hyung Jin; Leonardis, Ales.

Winter Conference on Applications of Computer Vision (WACV 2020). IEEE Computer Society Press, 2019.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Harvard

Chen, W, Duan, J, Basevi, H, Chang, HJ & Leonardis, A 2019, PointPoseNet: Point Pose Network for Robust 6D Object Pose Estimation. in Winter Conference on Applications of Computer Vision (WACV 2020). IEEE Computer Society Press, Winter Conference on Applications of Computer Vision (WACV 2020), Snowmass Village, Colorado, United States, 1/03/20.

APA

Chen, W., Duan, J., Basevi, H., Chang, H. J., & Leonardis, A. (Accepted/In press). PointPoseNet: Point Pose Network for Robust 6D Object Pose Estimation. In Winter Conference on Applications of Computer Vision (WACV 2020) IEEE Computer Society Press.

Vancouver

Chen W, Duan J, Basevi H, Chang HJ, Leonardis A. PointPoseNet: Point Pose Network for Robust 6D Object Pose Estimation. In Winter Conference on Applications of Computer Vision (WACV 2020). IEEE Computer Society Press. 2019

Author

Chen, Wei ; Duan, Jinming ; Basevi, Hector ; Chang, Hyung Jin ; Leonardis, Ales. / PointPoseNet : Point Pose Network for Robust 6D Object Pose Estimation. Winter Conference on Applications of Computer Vision (WACV 2020). IEEE Computer Society Press, 2019.

Bibtex

@inproceedings{d825386cdb5c4004932087b62635f7c5,
title = "PointPoseNet: Point Pose Network for Robust 6D Object Pose Estimation",
abstract = "In this paper, we propose a novel pipeline to estimate 6D object pose from RGB-D images of known objects present in complex scenes. The pipeline directly operates on raw point clouds extracted from RGB-D scans. Specifically, our method takes the point cloud as input and regresses the point-wise unit vectors pointing to the 3D keypoints. We then use these vectors to generate keypoint hypotheses from which the 6D object pose hypotheses are computed. Finally, we select the best 6D object pose from the hypotheses based on a proposed scoring mechanism with geometry constraints. Extensive experiments show that the proposed method is robust against the variety in object shape and appearance as well as occlusions between objects, and that our method outperforms the state-of-the-art methods on the LINEMOD and Occlusion LINEMOD datasets.",
author = "Wei Chen and Jinming Duan and Hector Basevi and Chang, {Hyung Jin} and Ales Leonardis",
year = "2019",
month = dec,
day = "9",
language = "English",
booktitle = "Winter Conference on Applications of Computer Vision (WACV 2020)",
publisher = "IEEE Computer Society Press",
note = "Winter Conference on Applications of Computer Vision (WACV 2020) ; Conference date: 01-03-2020 Through 05-03-2020",

}

RIS

TY - GEN

T1 - PointPoseNet

T2 - Winter Conference on Applications of Computer Vision (WACV 2020)

AU - Chen, Wei

AU - Duan, Jinming

AU - Basevi, Hector

AU - Chang, Hyung Jin

AU - Leonardis, Ales

PY - 2019/12/9

Y1 - 2019/12/9

N2 - In this paper, we propose a novel pipeline to estimate 6D object pose from RGB-D images of known objects present in complex scenes. The pipeline directly operates on raw point clouds extracted from RGB-D scans. Specifically, our method takes the point cloud as input and regresses the point-wise unit vectors pointing to the 3D keypoints. We then use these vectors to generate keypoint hypotheses from which the 6D object pose hypotheses are computed. Finally, we select the best 6D object pose from the hypotheses based on a proposed scoring mechanism with geometry constraints. Extensive experiments show that the proposed method is robust against the variety in object shape and appearance as well as occlusions between objects, and that our method outperforms the state-of-the-art methods on the LINEMOD and Occlusion LINEMOD datasets.

AB - In this paper, we propose a novel pipeline to estimate 6D object pose from RGB-D images of known objects present in complex scenes. The pipeline directly operates on raw point clouds extracted from RGB-D scans. Specifically, our method takes the point cloud as input and regresses the point-wise unit vectors pointing to the 3D keypoints. We then use these vectors to generate keypoint hypotheses from which the 6D object pose hypotheses are computed. Finally, we select the best 6D object pose from the hypotheses based on a proposed scoring mechanism with geometry constraints. Extensive experiments show that the proposed method is robust against the variety in object shape and appearance as well as occlusions between objects, and that our method outperforms the state-of-the-art methods on the LINEMOD and Occlusion LINEMOD datasets.

M3 - Conference contribution

BT - Winter Conference on Applications of Computer Vision (WACV 2020)

PB - IEEE Computer Society Press

Y2 - 1 March 2020 through 5 March 2020

ER -