Submitted by Jonas Schult.

Submission data

Full nameDualConvMesh-Net
DescriptionJoint Geodesic and Euclidean Convolutions on 3D Meshes
Publication titleDualConvMesh-Net: Joint Geodesic and Euclidean Convolutions on 3D Meshes
Publication authorsJonas Schult*, Francis Engelmann*, Theodora Kontogianni, Bastian Leibe
Publication venueCVPR 2020 [Oral]
Publication URLhttps://visualcomputinginstitute.github.io/dcm-net/
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)Python, PyTorch (Geometric)
HardwareTesla V100
Websitehttps://visualcomputinginstitute.github.io/dcm-net/
Source code or download URLhttps://github.com/VisualComputingInstitute/dcm-net
Submission creation date17 May, 2019
Last edited8 Apr, 2020

3D semantic label results

Infoavg ioubathtubbedbookshelfcabinetchaircountercurtaindeskdoorfloorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwallwindow
0.6580.7780.7020.8060.6190.8130.4680.6930.4940.5240.9410.4490.2980.5100.8210.6750.7270.5680.8260.8030.637