Uni6Dv2: Noise Elimination for 6D Pose Estimation

Mingshan Sun, Ye Zheng, Tianpeng Bao, Jianqiu Chen, Guoqiang Jin, Liwei Wu, Rui Zhao, Xiaoke Jiang
Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, PMLR 206:1832-1844, 2023.

Abstract

Uni6D is the first 6D pose estimation approach to employ a unified backbone network to extract features from both RGB and depth images. We discover that the principal reasons of Uni6D performance limitations are Instance-Outside and Instance-Inside noise. Uni6D’s simple pipeline design inherently introduces Instance-Outside noise from background pixels in the receptive field, while ignoring Instance-Inside noise in the input depth data. In this paper, we propose a two-step denoising approach for dealing with the aforementioned noise in Uni6D. To reduce noise from non-instance regions, an instance segmentation network is utilized in the first step to crop and mask the instance. A lightweight depth denoising module is proposed in the second step to calibrate the depth feature before feeding it into the pose regression network. Extensive experiments show that our Uni6Dv2 reliably and robustly eliminates noise, outperforming Uni6D without sacrificing too much inference efficiency. It also reduces the need for annotated real data that requires costly labeling.

Cite this Paper

BibTeX
@InProceedings{pmlr-v206-sun23b, title = {Uni6Dv2: Noise Elimination for 6D Pose Estimation}, author = {Sun, Mingshan and Zheng, Ye and Bao, Tianpeng and Chen, Jianqiu and Jin, Guoqiang and Wu, Liwei and Zhao, Rui and Jiang, Xiaoke}, booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics}, pages = {1832--1844}, year = {2023}, editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem}, volume = {206}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v206/sun23b/sun23b.pdf}, url = {https://proceedings.mlr.press/v206/sun23b.html}, abstract = {Uni6D is the first 6D pose estimation approach to employ a unified backbone network to extract features from both RGB and depth images. We discover that the principal reasons of Uni6D performance limitations are Instance-Outside and Instance-Inside noise. Uni6D’s simple pipeline design inherently introduces Instance-Outside noise from background pixels in the receptive field, while ignoring Instance-Inside noise in the input depth data. In this paper, we propose a two-step denoising approach for dealing with the aforementioned noise in Uni6D. To reduce noise from non-instance regions, an instance segmentation network is utilized in the first step to crop and mask the instance. A lightweight depth denoising module is proposed in the second step to calibrate the depth feature before feeding it into the pose regression network. Extensive experiments show that our Uni6Dv2 reliably and robustly eliminates noise, outperforming Uni6D without sacrificing too much inference efficiency. It also reduces the need for annotated real data that requires costly labeling.} }
Endnote
%0 Conference Paper %T Uni6Dv2: Noise Elimination for 6D Pose Estimation %A Mingshan Sun %A Ye Zheng %A Tianpeng Bao %A Jianqiu Chen %A Guoqiang Jin %A Liwei Wu %A Rui Zhao %A Xiaoke Jiang %B Proceedings of The 26th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2023 %E Francisco Ruiz %E Jennifer Dy %E Jan-Willem van de Meent %F pmlr-v206-sun23b %I PMLR %P 1832--1844 %U https://proceedings.mlr.press/v206/sun23b.html %V 206 %X Uni6D is the first 6D pose estimation approach to employ a unified backbone network to extract features from both RGB and depth images. We discover that the principal reasons of Uni6D performance limitations are Instance-Outside and Instance-Inside noise. Uni6D’s simple pipeline design inherently introduces Instance-Outside noise from background pixels in the receptive field, while ignoring Instance-Inside noise in the input depth data. In this paper, we propose a two-step denoising approach for dealing with the aforementioned noise in Uni6D. To reduce noise from non-instance regions, an instance segmentation network is utilized in the first step to crop and mask the instance. A lightweight depth denoising module is proposed in the second step to calibrate the depth feature before feeding it into the pose regression network. Extensive experiments show that our Uni6Dv2 reliably and robustly eliminates noise, outperforming Uni6D without sacrificing too much inference efficiency. It also reduces the need for annotated real data that requires costly labeling.
APA
Sun, M., Zheng, Y., Bao, T., Chen, J., Jin, G., Wu, L., Zhao, R. & Jiang, X.. (2023). Uni6Dv2: Noise Elimination for 6D Pose Estimation. Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 206:1832-1844 Available from https://proceedings.mlr.press/v206/sun23b.html.

Related Material