Year
Month
(Preprint) Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks
Zhihao Liang ¹ ², Zhihao Li ³, Songcen Xu 许松岑 ³, Mingkui Tan 谭明奎 ¹, Kui Jia 贾奎 ¹ ⁴ ⁵
¹ South China University of Technology
华南理工大学
² DexForce Technology Co., Ltd.
跨维(广州)智能科技有限公司
³ Noah’s Ark Lab, Huawei Technologies
华为诺亚方舟实验室
⁴ Pazhou Laboratory
琶洲实验室 (人工智能与数字经济广东省实验室)
⁵ Peng Cheng Laboratory
鹏城实验室
arXiv, 2021-08-17
Abstract

Instance 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 semantic features of superpoints, and which will be traversed and split at intermediate tree nodes for proposals of object instances. We also design in SSTNet a refinement module, termed CliqueNet, to prune superpoints that may be wrongly grouped into instance proposals.

Experiments on the benchmarks of ScanNet and S3DIS show the efficacy of our proposed method. At the time of submission, SSTNet ranks top on the ScanNet (V2) leaderboard, with 2% higher of mAP than the second best method.
Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks_1
Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks_2
Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks_3
Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks_4
  • A novel approach towards robust construction of physical colors on lithium niobate crystal
  • Quanxin Yang, Menghan Yu, Zhixiang Chen, Siwen Ai, Ulrich Kentsch, Shengqiang Zhou, Yuechen Jia, Feng Chen, Hongliang Liu
  • Opto-Electronic Advances
  • 2025-01-22
  • Multi-photon neuron embedded bionic skin for high-precision complex texture and object reconstruction perception research
  • Hongyu Zhou, Chao Zhang, Hengchang Nong, Junjie Weng, Dongying Wang, Yang Yu, Jianfa Zhang, Chaofan Zhang, Jinran Yu, Zhaojian Zhang, Huan Chen, Zhenrong Zhang, Junbo Yang
  • Opto-Electronic Advances
  • 2025-01-22
  • Single-beam optical trap-based surface-enhanced raman scattering optofluidic molecular fingerprint spectroscopy detection system
  • Ning Sun, Yuan Gan, Yujie Wu, Xing Wang, Shen Shen, Yong Zhu, Jie Zhang
  • Opto-Electronic Advances
  • 2025-01-22
  • High-frequency enhanced ultrafast compressed active photography
  • Yizhao Meng, Yu Lu, Pengfei Zhang, Yi Liu, Fei Yin, Lin Kai, Qing Yang, Feng Chen
  • Opto-Electronic Advances
  • 2025-01-15
  • Efficient generation of vectorial terahertz beams using surface-wave excited metasurfaces
  • Zhuo Wang, Weikang Pan, Yu He, Zhiyan Zhu, Xiangyu Jin, Muhan Liu, Shaojie Ma, Qiong He, Shulin Sun, Lei Zhou
  • Opto-Electronic Science
  • 2025-01-15
  • High-efficiency RGB achromatic liquid crystal diffractive optical elements
  • Yuqiang Ding, Xiaojin Huang, Yongziyan Ma, Yan Li, Shin-Tson Wu
  • Opto-Electronic Advances
  • 2025-01-07
  • On-chip light control of semiconductor optoelectronic devices using integrated metasurfaces
  • Cheng-Long Zheng, Pei-Nan Ni, Yi-Yang Xie, Patrice Genevet
  • Opto-Electronic Advances
  • 2025-01-07
  • Ferroelectric domain engineering of lithium niobate
  • Jackson J. Chakkoria, Aditya Dubey, Arnan Mitchell, Andreas Boes
  • Opto-Electronic Advances
  • 2025-01-03
  • Smart reconfigurable metadevices made of shape memory alloy metamaterials
  • Shiqiang Zhao, Yuancheng Fan, Ruisheng Yang, Zhehao Ye, Fuli Zhang, Chen Wang, Weijia Luo, Yongzheng Wen, Ji Zhou
  • Opto-Electronic Advances
  • 2025-01-03
  • Direct detection with an optimal transfer function: toward the electrical spectral efficiency of coherent homodyne detection
  • Xingfeng Li, Jingchi Li, Xiong Ni, Hudi Liu, Qunbi Zhuge, Haoshuo Chen, William Shieh, Yikai Su
  • Opto-Electronic Science
  • 2024-12-24
  • Enhanced amplified spontaneous emission via splitted strong coupling mode in large-area plasmonic cone lattices
  • Jiazhi Yuan, Jiang Hu, Yan Zheng, Hao Wei, Jiamin Xiao, Yi Wang, Xuchao Zhao, Ye Xiang, Yong Lei, Wenxin Wang
  • Opto-Electronic Science
  • 2024-12-19
  • High-resolution tumor marker detection based on microwave photonics demodulated dual wavelength fiber laser sensor
  • Jie Hu, Weihao Lin, Liyang Shao, Chenlong Xue, Fang Zhao, Dongrui Xiao, Yang Ran, Yue Meng, Panpan He, Zhiguang Yu, Jinna Chen, Perry Ping Shum
  • Opto-Electronic Advances
  • 2024-12-16



  • Integrated Sensing and Communications: Towards Dual-functional Wireless Networks for 6G and Beyond                                Successful New-entry Prediction for Multi-Party Online Conversations via Latent Topics and Discourse Modeling
    About
    |
    Contact
    |
    Copyright © PubCard