Submitted by weiguang Zhao.

Submission data

Full nameDivide and Conquer: 3D Instance Segmentation With Point-Wise Binarization
DescriptionBinarization and classification of each point in the point cloud
Publication titleDivide and Conquer: 3D Instance Segmentation With Point-Wise Binarization
Publication authorsW.Zhao, Y.Yan, C.Yang, J.Ye,X.Yang,K.Huang
Publication venueICCV 2023
Publication URLhttps://openaccess.thecvf.com/content/ICCV2023/html/Zhao_Divide_and_Conquer_3D_Point_Cloud_Instance_Segmentation_With_Point-Wise_ICCV_2023_paper.html
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)C++ Cuda python pytorch
Hardware3090
Submission creation date25 Nov, 2021
Last edited10 Oct, 2023

3D semantic instance results



Infoavg ap 50%bathtubbedbookshelfcabinetchaircountercurtaindeskdoorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwindow
permissive0.7471.0000.8180.8370.7130.8440.4570.6470.7110.6140.6170.6570.6501.0000.6920.8220.7651.0000.595