Submitted by Hugues THOMAS.

Submission data

Full nameKPConv: Flexible and Deformable Convolution for Point Clouds
DescriptionKPConv Segmentation Network for Point Clouds
Publication titleKPConv: Flexible and Deformable Convolution for Point Clouds
Publication authorsH. Thomas, C. Qi, J. Deschaud, B. Marcotegui, F. Goulette, L. Guibas.
Publication venueICCV 2019
Publication URLhttps://arxiv.org/abs/1904.08889
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)python
HardwareIntel(R) Core(TM) i7-3770 CPU @ 3.40GHz, 32GB RAM, Nvidia GeForce GTX 1080 Ti
Websitehttps://github.com/HuguesTHOMAS/KPConv
Submission creation date14 Aug, 2018
Last edited19 Aug, 2019

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.6840.8470.7580.7840.6470.8140.4730.7720.6050.5940.9350.4500.1810.5870.8050.6900.7850.6140.8820.8190.632