Submitted by Chen Liu.

Submission data

Full nameMASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation
DescriptionWe propose a new approach for 3D instance segmentation based on sparse convolution and point affinity prediction, which indicates the likelihood of two points belonging to the same instance. The proposed network, built upon
submanifold sparse convolution [3], processes a voxelized
point cloud and predicts semantic scores for each occupied
voxel as well as the affinity between neighboring voxels
at different scales. A simple yet effective clustering algorithm segments points into instances based on the predicted
affinity and the mesh topology. The semantic for each instance is determined by the semantic prediction. Experiments show that our method outperforms the state-of-theart instance segmentation methods by a large margin on the
widely used ScanNet benchmark [2]. We share our code
publicly at https://github.com/art-programmer/MASC.
Publication titleMASC: Multi-scale Affinity with Sparse Convolution for 3D Instance Segmentation
Publication authorsChen Liu, Yasutaka Furukawa
Publication URLhttps://arxiv.org/abs/1902.04478
Input Data TypesUses Color,Uses Geometry        Uses 3D
Programming language(s)PyTorch
HardwareTitanX
Websitehttps://github.com/art-programmer/MASC/blob/master/README.md
Source code or download URLhttps://github.com/art-programmer/MASC/blob/master/README.md
Submission creation date17 Jan, 2019
Last edited13 Feb, 2019

3D semantic instance results



Infoavg ap 50%bathtubbedbookshelfcabinetchaircountercurtaindeskdoorotherfurniturepicturerefrigeratorshower curtainsinksofatabletoiletwindow
permissive0.4470.5280.5550.3810.3820.6330.0020.5090.2600.3610.4320.3270.4510.5710.3670.6390.3860.9800.276