Submitted by Zhihao Liang.

Submission data

Full nameSemantic Superpoint Tree Network
DescriptionInstance segmentation in 3D scenes is fundamental in many applications of scene understanding. It is yet challenging due to the compound factors of data irregularity and uncertainty in the numbers of instances. State-of-the-art methods largely rely on a general pipeline that first learns point-wise features discriminative at semantic and instance levels, followed by a separate step of point grouping for proposing object instances. While promising, they have the shortcomings that (1) the second step is not supervised by the main objective of instance segmentation, and (2) their point-wise feature learning and grouping are less effective to deal with data irregularities, possibly resulting in fragmented segmentations. To address these issues, we propose in this work an end-to-end solution of Semantic Superpoint Tree Network (SSTNet) for proposing object instances from scene points. Key in SSTNet is an intermediate, semantic superpoint tree (SST), which is constructed based on the learned
Publication titleInstance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks
Publication authorsZhihao Liang, Zhihao Li, Songcen Xu, Mingkui Tan, Kui Jia
Publication venueICCV2021
Publication URLhttps://arxiv.org/pdf/2108.07478.pdf
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)Python with CUDA
HardwareRTX 3090
Websitehttps://github.com/Gorilla-Lab-SCUT/SSTNet
Source code or download URLhttps://github.com/Gorilla-Lab-SCUT/SSTNet
Submission creation date4 Jan, 2021
Last edited19 Aug, 2021

3D semantic instance results



Infoavg ap 50%bathtubbedbookshelfcabinetchaircountercurtaindeskdoorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwindow
permissive0.6981.0000.6970.8880.5560.8030.3870.6260.4170.5560.5850.7020.6001.0000.8240.7200.6921.0000.509