Submitted by Angela Dai.

Submission data

Full nameScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes
Description2D projections of 3D semantic segmentation predictions
Publication titleScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes
Publication authorsAngela Dai, Angel X. Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, Matthias Nießner
Publication venueCVPR'17
Publication URLhttp://www.scan-net.org/
Input Data TypesUses Geometry        Uses 3D
Programming language(s)pytorch
HardwareGTX 1080
Websitehttp://www.scan-net.org/
Source code or download URLhttps://github.com/scannet/scannet
Submission creation date13 Jul, 2018
Last edited20 Jul, 2018

2D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
permissive0.3300.2930.5210.6570.3610.1610.2500.0040.4400.1830.8360.1250.0600.3190.1320.4170.4120.3440.5410.4270.109