Submitted by Armen Avetisyan.

Submission data

Full nameScan2CAD: Correspondence Prediction and 9DoF Alignment
Publication titleScan2CAD: Learning CAD Model Alignment in RGB-D Scans
Publication authorsArmen Avetisyan, Manuel Dahnert, Angela Dai, Manolis Savva, Angel X. Chang, Matthias Niessner
Publication venuearXiv
Publication URLhttps://arxiv.org/pdf/1811.11187.pdf
Input Data TypesUses Geometry        Uses 3D
Programming language(s)python3 with pytorch, c++
HardwareNvidia 1080 GTX
Websitehttps://github.com/skanti/Scan2CAD
Source code or download URLhttps://github.com/skanti/Scan2CAD
Submission creation date14 Jan, 2019
Last edited14 Jan, 2019

CAD Alignment

Infoavg (per CAD)avg (per class)bathtubbookshelfcabinetchairdisplayothersofatabletrashbin
permissive0.2880.2890.2500.2810.3080.3970.2030.1430.5110.3150.196