Submitted by Wenxuan Wu.

Submission data

Full namePointConv: Deep Convolutional Networks on 3D Point Clouds
DescriptionPointConv: Deep Convolutional Networks on 3D Point Clouds
Publication titlePointConv: Deep Convolutional Networks on 3D Point Clouds
Publication authorsWenxuan Wu, Zhongang Qi, Li Fuxin
Publication venueCVPR 2019
Publication URLhttps://arxiv.org/abs/1811.07246
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)python with CUDA
HardwareTitan RTX
Websitehttp://web.engr.oregonstate.edu/~wuwen/pointconv.html
Source code or download URLhttps://github.com/DylanWusee/pointconv
Submission creation date8 Nov, 2018
Last edited11 Oct, 2019

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
permissive0.6660.7810.7590.6990.6440.8220.4750.7790.5640.5040.9530.4280.2030.5860.7540.6610.7530.5880.9020.8130.642