Submitted by Min Zhong.

Submission data

Full nameMaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation
DescriptionWe propose a novel framework to group and refine the 3D instances. In practice, we first learn an offset vector for each point and shift it to its predicted instance center. To better group these points, we propose a Hierarchical Point Grouping algorithm to merge the centrally aggregated points progressively. All points are grouped into small clusters, which further gradually undergo another clustering procedure to merge into larger groups. These multi-scale groups are exploited for instance prediction, which is beneficial for predicting instances with different scales. In addition, a novel MaskScoreNet is developed to produce binary point masks of these groups for further refining the segmentation results.
Publication titleMaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation
Publication authorsMin Zhong, Xinghao Chen, Xiaokang Chen, Gang Zeng, Yunhe Wang
Publication venueICME 2022
Publication URLhttps://arxiv.org/abs/2203.14662
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)python
Hardware1080ti
Submission creation date7 Nov, 2020
Last edited29 Mar, 2022

3D semantic instance results



Infoavg ap 25%bathtubbedbookshelfcabinetchaircountercurtaindeskdoorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwindow
0.7921.0000.9680.8120.7660.8640.4600.8150.8880.5980.6510.6390.6000.9180.9410.8960.7211.0000.723