3D Instance Segmentation On S3Dis

Metrics

AP@50

mAP

mCov

mPrec

mRec

mWCov

Results

Performance results of various models on this benchmark

Model Name	AP@50	mAP	mCov	mPrec	mRec	mWCov	Paper Title	Repository
SoftGroup	68.9	54.4	69.3	75.3	69.8	71.7	SoftGroup for 3D Instance Segmentation on Point Clouds
PointGroup	64.0	-	-	69.6	69.2	-	PointGroup: Dual-Set Point Grouping for 3D Instance Segmentation
ISBNet	70.5	60.8	74.9	77.5	77.1	76.8	ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution
3D-MPA	-	-	-	66.7	64.1	-	3D-MPA: Multi Proposal Aggregation for 3D Semantic Instance Segmentation
SPFormer	69.2	-	-	74.0	71.1	-	Superpoint Transformer for 3D Scene Instance Segmentation
MSTA3D	70.0	-	-	80.6	70.1	-	MSTA3D: Multi-scale Twin-attention for 3D Instance Segmentation	-
OneFormer3D	75.8	63.0	-	82.3	74.1	-	OneFormer3D: One Transformer for Unified Point Cloud Segmentation
3D-BoNet	-	-	-	65.6	47.6	-	Learning Object Bounding Boxes for 3D Instance Segmentation on Point Clouds
JSNet	-	-	54.1	66.9	53.9	58	JSNet: Joint Instance and Semantic Segmentation of 3D Point Clouds
HAIS	-	-	67.0	73.2	69.4	70.4	Hierarchical Aggregation for 3D Instance Segmentation
PartNet	-	-	-	-	43.4%	-	PartNet: A Recursive Part Decomposition Network for Fine-grained and Hierarchical Shape Segmentation	-
PBNet	70.6	59.5	-	-	-	-	Divide and Conquer: 3D Point Cloud Instance Segmentation With Point-Wise Binarization
SASO	-	-	54.5	64.2	50.8	58.3	SASO: Joint 3D Semantic-Instance Segmentation via Multi-scale Semantic Association and Salient Point Clustering Optimization	-
TD3D	70.4	58.1	-	-	-	-	Top-Down Beats Bottom-Up in 3D Instance Segmentation
MaskGroup	69.9	-	-	66.6	69.6	-	MaskGroup: Hierarchical Point Grouping and Masking for 3D Instance Segmentation	-
ASIS	-	-	-	63.6	47.5	-	Associatively Segmenting Instances and Semantics in Point Clouds
PointCNN	-	-	-	-	-	-	PointCNN: Convolution On $mathcal{X}$-Transformed Points
Mask3D	75.5	64.5	-	-	-	-	Mask3D: Mask Transformer for 3D Semantic Instance Segmentation
GICN	-	-	-	68.5	50.8	-	Learning Gaussian Instance Segmentation in Point Clouds
DKNet	-	-	70.3	75.3	71.1	72.8	3D Instances as 1D Kernels

0 of 21 row(s) selected.