Semantic Segmentation on ScanNet

Metric: val mIoU (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	val mIoU▼	Extra Data	Paper	Date↕	Code
1	DITR	80.5	No	DINO in the Room: Leveraging 2D Foundation Model...	2025-03-24	Code
2	Sonata + PTv3	79.4	Yes	Sonata: Self-Supervised Learning of Reliable Poi...	2025-03-20	Code
3	PTv3 ARKit LabelMaker	79.1	Yes	ARKit LabelMaker: A New Scale for Indoor 3D Scen...	2024-10-17	Code
4	PTv3 + PPT	78.6	Yes	Point Transformer V3: Simpler, Faster, Stronger	2023-12-15	Code
5	BFANet	78	No	BFANet: Revisiting 3D Semantic Segmentation with...	2025-03-16	Code
6	ODIN	77.8	No	ODIN: A Single Model for 2D and 3D Segmentation	2024-01-04	Code
7	Pamba	77.6	No	Pamba: Enhancing Global Interaction in Point Clo...	2024-06-25	-
8	Swin3D-L	77.5	Yes	Swin3D: A Pretrained Transformer Backbone for 3D...	2023-04-14	Code
9	PonderV2 + SparseUNet	77	Yes	PonderV2: Pave the Way for 3D Foundation Model w...	2023-10-12	Code
10	Serialized Piont Mamba	76.8	No	Serialized Point Mamba: A Serialized Point Cloud...	2024-07-17	-
11	OneFormer3D	76.6	No	OneFormer3D: One Transformer for Unified Point C...	2023-11-24	Code
12	PPT + SparseUNet	76.4	Yes	Towards Large-scale 3D Representation Learning w...	2023-08-18	Code
13	KPConvX-L	76.3	No	KPConvX: Modernizing Kernel Point Convolution wi...	2024-05-21	Code
14	OA-CNNs	76.1	No	OA-CNNs: Omni-Adaptive Sparse CNNs for 3D Semant...	2024-03-21	Code
15	AVS-Net	76.1	No	AVS-Net: Point Sampling with Adaptive Voxel Size...	2024-02-27	Code
16	DeLA	75.9	No	Decoupled Local Aggregation for Point Cloud Lear...	2023-08-31	Code
17	OctFormer	75.7	No	OctFormer: Octree-based Transformers for 3D Poin...	2023-05-04	Code
18	LSK3DNet	75.7	No	LSK3DNet: Towards Effective and Efficient 3D Per...	2024-03-22	Code
19	MSC + SparseUNet	75.5	Yes	Masked Scene Contrast: A Scalable Framework for ...	2023-03-24	Code
20	PointHR	75.4	No	PointHR: Exploring High-Resolution Architectures...	2023-10-11	Code
21	PTv2	75.4	No	Point Transformer V2: Grouped Vector Attention a...	2022-10-11	Code
22	EQ-Net	75.3	No	A Unified Query-based Paradigm for Point Cloud U...	2022-03-02	Code
23	StratifiedFormer	74.3	No	Stratified Transformer for 3D Point Cloud Segmen...	2022-03-28	Code
24	O-CNN	74	No	O-CNN: Octree-based Convolutional Neural Network...	2017-12-05	Code
25	BPNet	73.9	No	Bidirectional Projection Network for Cross Dimen...	2021-03-26	Code
26	Mix3D	73.6	No	Mix3D: Out-of-Context Data Augmentation for 3D S...	2021-10-05	Code
27	MinkowskiNet	72.2	No	4D Spatio-Temporal ConvNets: Minkowski Convoluti...	2019-04-18	Code
28	SparseConvNet	69.3	No	3D Semantic Segmentation with Submanifold Sparse...	2017-11-28	Code
29	KpConv	69.2	No	KPConv: Flexible and Deformable Convolution for ...	2019-04-18	Code
30	PanopticNDT (10cm)	68.39	No	PanopticNDT: Efficient and Robust Panoptic Mapping	2023-09-24	Code
31	PointConv	61	No	PointConv: Deep Convolutional Networks on 3D Poi...	2018-11-17	Code
32	PointNet++	53.5	No	PointNet++: Deep Hierarchical Feature Learning o...	2017-06-07	Code

#1DITRSOTA
80.5
val mIoU· 2025-03-24
DINO in the Room: Leveraging 2D Foundation Models for 3D Segmentation Code
#2Sonata + PTv3SOTA
79.4
val mIoU· Extra Data· 2025-03-20
Sonata: Self-Supervised Learning of Reliable Point Representations Code
#3PTv3 ARKit LabelMakerSOTA
79.1
val mIoU· Extra Data· 2024-10-17
ARKit LabelMaker: A New Scale for Indoor 3D Scene Understanding Code
#4PTv3 + PPTSOTA
78.6
val mIoU· Extra Data· 2023-12-15
Point Transformer V3: Simpler, Faster, Stronger Code
#5BFANet
78
val mIoU· 2025-03-16
BFANet: Revisiting 3D Semantic Segmentation with Boundary Feature Analysis Code
#6ODIN
77.8
val mIoU· 2024-01-04
ODIN: A Single Model for 2D and 3D Segmentation Code
#7Pamba
77.6
val mIoU· 2024-06-25
Pamba: Enhancing Global Interaction in Point Clouds via State Space Model
#8Swin3D-LSOTA
77.5
val mIoU· Extra Data· 2023-04-14
Swin3D: A Pretrained Transformer Backbone for 3D Indoor Scene Understanding Code
#9PonderV2 + SparseUNet
77
val mIoU· Extra Data· 2023-10-12
PonderV2: Pave the Way for 3D Foundation Model with A Universal Pre-training Paradigm Code
#10Serialized Piont Mamba
76.8
val mIoU· 2024-07-17
Serialized Point Mamba: A Serialized Point Cloud Mamba Segmentation Model
#11OneFormer3D
76.6
val mIoU· 2023-11-24
OneFormer3D: One Transformer for Unified Point Cloud Segmentation Code
#12PPT + SparseUNet
76.4
val mIoU· Extra Data· 2023-08-18
Towards Large-scale 3D Representation Learning with Multi-dataset Point Prompt Training Code
#13KPConvX-L
76.3
val mIoU· 2024-05-21
KPConvX: Modernizing Kernel Point Convolution with Kernel Attention Code
#14OA-CNNs
76.1
val mIoU· 2024-03-21
OA-CNNs: Omni-Adaptive Sparse CNNs for 3D Semantic Segmentation Code
#15AVS-Net
76.1
val mIoU· 2024-02-27
AVS-Net: Point Sampling with Adaptive Voxel Size for 3D Scene Understanding Code
#16DeLA
75.9
val mIoU· 2023-08-31
Decoupled Local Aggregation for Point Cloud Learning Code
#17OctFormer
75.7
val mIoU· 2023-05-04
OctFormer: Octree-based Transformers for 3D Point Clouds Code
#18LSK3DNet
75.7
val mIoU· 2024-03-22
LSK3DNet: Towards Effective and Efficient 3D Perception with Large Sparse Kernels Code
#19MSC + SparseUNetSOTA
75.5
val mIoU· Extra Data· 2023-03-24
Masked Scene Contrast: A Scalable Framework for Unsupervised 3D Representation Learning Code
#20PointHR
75.4
val mIoU· 2023-10-11
PointHR: Exploring High-Resolution Architectures for 3D Point Cloud Segmentation Code
#21PTv2SOTA
75.4
val mIoU· 2022-10-11
Point Transformer V2: Grouped Vector Attention and Partition-based Pooling Code
#22EQ-NetSOTA
75.3
val mIoU· 2022-03-02
A Unified Query-based Paradigm for Point Cloud Understanding Code
#23StratifiedFormer
74.3
val mIoU· 2022-03-28
Stratified Transformer for 3D Point Cloud Segmentation Code
#24O-CNNSOTA
74
val mIoU· 2017-12-05
O-CNN: Octree-based Convolutional Neural Networks for 3D Shape Analysis Code
#25BPNet
73.9
val mIoU· 2021-03-26
Bidirectional Projection Network for Cross Dimension Scene Understanding Code
#26Mix3D
73.6
val mIoU· 2021-10-05
Mix3D: Out-of-Context Data Augmentation for 3D Scenes Code
#27MinkowskiNet
72.2
val mIoU· 2019-04-18
4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks Code
#28SparseConvNetSOTA
69.3
val mIoU· 2017-11-28
3D Semantic Segmentation with Submanifold Sparse Convolutional Networks Code
#29KpConv
69.2
val mIoU· 2019-04-18
KPConv: Flexible and Deformable Convolution for Point Clouds Code
#30PanopticNDT (10cm)
68.39
val mIoU· 2023-09-24
PanopticNDT: Efficient and Robust Panoptic Mapping Code
#31PointConv
61
val mIoU· 2018-11-17
PointConv: Deep Convolutional Networks on 3D Point Clouds Code
#32PointNet++SOTA
53.5
val mIoU· 2017-06-07
PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space Code