3D Point Cloud Reconstruction on ScanObjectNN

Metric: OBJ-ONLY (OA) (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	OBJ-ONLY (OA)▼	Extra Data	Paper	Date↕	Code
1	PointGST	97.76	Yes	Parameter-Efficient Fine-Tuning in Spectral Doma...	2024-10-10	Code
2	ReCon++	97.59	Yes	ShapeLLM: Universal 3D Object Understanding for ...	2024-02-27	Code
3	PointGPT	96.6	Yes	PointGPT: Auto-regressively Generative Pre-train...	2023-05-19	Code
4	3D-JEPA	94.49	Yes	3D-JEPA: A Joint Embedding Predictive Architectu...	2024-09-24	-
5	AsymDSD-B* (no voting)	94.32	Yes	Asymmetric Dual Self-Distillation for 3D Self-Su...	2025-06-26	Code
6	PCP-MAE	94.32	No	PCP-MAE: Learning to Predict Centers for Point M...	2024-08-16	Code
7	ReCon	93.8	Yes	Contrast with Reconstruct: Contrastive 3D Repres...	2023-02-05	Code
8	Point-RAE (no voting)	93.63	Yes	Regress Before Construct: Regress Autoencoder fo...	2023-09-25	Code
9	ReCon (no voting)	93.29	Yes	Contrast with Reconstruct: Contrastive 3D Repres...	2023-02-05	Code
10	Point-FEMAE	93.29	Yes	Towards Compact 3D Representations via Point Fea...	2023-12-17	Code
11	ReCon+PPT	93.28	No	Positional Prompt Tuning for Efficient 3D Repres...	2024-08-21	Code
12	Mamba3D	92.43	No	Mamba3D: Enhancing Local Features for 3D Point C...	2024-04-23	Code
13	OTMae3D	92.3	Yes	-	-	Code
14	MVTN	92.3	No	MVTN: Multi-View Transformation Network for 3D S...	2020-11-26	Code
15	Mamba3D (no voting)	92.08	No	Mamba3D: Enhancing Local Features for 3D Point C...	2024-04-23	Code
16	AsymDSD-S (no voting)	91.91	Yes	Asymmetric Dual Self-Distillation for 3D Self-Su...	2025-06-26	Code
17	ACT (no voting)	91.91	Yes	Autoencoders as Cross-Modal Teachers: Can Pretra...	2022-12-16	Code
18	I2P-MAE (no voting)	91.57	Yes	Learning 3D Representations from 2D Pre-trained ...	2022-12-13	Code
19	point2vec	90.4	Yes	Point2Vec for Self-Supervised Representation Lea...	2023-03-29	Code
20	ExpPoint-MAE	90.02	Yes	ExpPoint-MAE: Better interpretability and perfor...	2023-06-19	Code
21	Point-LGMask	89.3	Yes	-	-	Code
22	Point-M2AE	88.81	Yes	Point-M2AE: Multi-scale Masked Autoencoders for ...	2022-05-28	Code
23	Point-MAE	88.29	Yes	Masked Autoencoders for Point Cloud Self-supervi...	2022-03-13	Code
24	Point-BERT	88.12	Yes	Point-BERT: Pre-training 3D Point Cloud Transfor...	2021-11-29	Code
25	DGCNN	86.2	No	Dynamic Graph CNN for Learning on Point Clouds	2018-01-24	Code
26	PointCNN	85.5	No	PointCNN: Convolution On $\mathcal{X}$-Transform...	2018-01-23	Code
27	PointNet++	84.3	No	PointNet++: Deep Hierarchical Feature Learning o...	2017-06-07	Code

#1PointGSTSOTA
97.76
OBJ-ONLY (OA)· Extra Data· 2024-10-10
Parameter-Efficient Fine-Tuning in Spectral Domain for Point Cloud Learning Code
#2ReCon++SOTA
97.59
OBJ-ONLY (OA)· Extra Data· 2024-02-27
ShapeLLM: Universal 3D Object Understanding for Embodied Interaction Code
#3PointGPTSOTA
96.6
OBJ-ONLY (OA)· Extra Data· 2023-05-19
PointGPT: Auto-regressively Generative Pre-training from Point Clouds Code
#43D-JEPA
94.49
OBJ-ONLY (OA)· Extra Data· 2024-09-24
3D-JEPA: A Joint Embedding Predictive Architecture for 3D Self-Supervised Representation Learning
#5AsymDSD-B* (no voting)
94.32
OBJ-ONLY (OA)· Extra Data· 2025-06-26
Asymmetric Dual Self-Distillation for 3D Self-Supervised Representation Learning Code
#6PCP-MAE
94.32
OBJ-ONLY (OA)· 2024-08-16
PCP-MAE: Learning to Predict Centers for Point Masked Autoencoders Code
#7ReConSOTA
93.8
OBJ-ONLY (OA)· Extra Data· 2023-02-05
Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining Code
#8Point-RAE (no voting)
93.63
OBJ-ONLY (OA)· Extra Data· 2023-09-25
Regress Before Construct: Regress Autoencoder for Point Cloud Self-supervised Learning Code
#9ReCon (no voting)
93.29
OBJ-ONLY (OA)· Extra Data· 2023-02-05
Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining Code
#10Point-FEMAE
93.29
OBJ-ONLY (OA)· Extra Data· 2023-12-17
Towards Compact 3D Representations via Point Feature Enhancement Masked Autoencoders Code
#11ReCon+PPT
93.28
OBJ-ONLY (OA)· 2024-08-21
Positional Prompt Tuning for Efficient 3D Representation Learning Code
#12Mamba3D
92.43
OBJ-ONLY (OA)· 2024-04-23
Mamba3D: Enhancing Local Features for 3D Point Cloud Analysis via State Space Model Code
#13OTMae3D
92.3
OBJ-ONLY (OA)· Extra Data
No paperCode
#14MVTNSOTA
92.3
OBJ-ONLY (OA)· 2020-11-26
MVTN: Multi-View Transformation Network for 3D Shape Recognition Code
#15Mamba3D (no voting)
92.08
OBJ-ONLY (OA)· 2024-04-23
Mamba3D: Enhancing Local Features for 3D Point Cloud Analysis via State Space Model Code
#16AsymDSD-S (no voting)
91.91
OBJ-ONLY (OA)· Extra Data· 2025-06-26
Asymmetric Dual Self-Distillation for 3D Self-Supervised Representation Learning Code
#17ACT (no voting)
91.91
OBJ-ONLY (OA)· Extra Data· 2022-12-16
Autoencoders as Cross-Modal Teachers: Can Pretrained 2D Image Transformers Help 3D Representation Learning?Code
#18I2P-MAE (no voting)
91.57
OBJ-ONLY (OA)· Extra Data· 2022-12-13
Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoders Code
#19point2vec
90.4
OBJ-ONLY (OA)· Extra Data· 2023-03-29
Point2Vec for Self-Supervised Representation Learning on Point Clouds Code
#20ExpPoint-MAE
90.02
OBJ-ONLY (OA)· Extra Data· 2023-06-19
ExpPoint-MAE: Better interpretability and performance for self-supervised point cloud transformers Code
#21Point-LGMask
89.3
OBJ-ONLY (OA)· Extra Data
No paperCode
#22Point-M2AE
88.81
OBJ-ONLY (OA)· Extra Data· 2022-05-28
Point-M2AE: Multi-scale Masked Autoencoders for Hierarchical Point Cloud Pre-training Code
#23Point-MAE
88.29
OBJ-ONLY (OA)· Extra Data· 2022-03-13
Masked Autoencoders for Point Cloud Self-supervised Learning Code
#24Point-BERT
88.12
OBJ-ONLY (OA)· Extra Data· 2021-11-29
Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modeling Code
#25DGCNNSOTA
86.2
OBJ-ONLY (OA)· 2018-01-24
Dynamic Graph CNN for Learning on Point Clouds Code
#26PointCNNSOTA
85.5
OBJ-ONLY (OA)· 2018-01-23
PointCNN: Convolution On $\mathcal{X}$-Transformed Points Code
#27PointNet++SOTA
84.3
OBJ-ONLY (OA)· 2017-06-07
PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space Code