paper-with-me

3D Point Cloud Classification 벤치마크

3D Point Cloud Classification on ScanObjectNN

231개 결과 · ⬇ CSV · JSON

Overall Accuracy

68.2 75.45 82.7 89.95 97.2 2015-12 2026-11 PointNeXt+GAM — 88.4 (2015-12-14) PointNeXt+GAM — 88.4 (2015-12-14) PointNeXt+GAM — 88.4 (2015-12-14) PointNet — 68.2 (2016-12-02) PointNet — 68.2 (2016-12-02) PointNet — 68.2 (2016-12-02) PointNet++ — 77.9 (2017-06-07) PointNet++ — 77.9 (2017-06-07) PointNet++ — 77.9 (2017-06-07) PointCNN — 78.5 (2018-01-23) PointCNN — 78.5 (2018-01-23) PointCNN — 78.5 (2018-01-23) DGCNN — 78.1 (2018-01-24) DGCNN — 78.1 (2018-01-24) DGCNN — 78.1 (2018-01-24) SpiderCNN — 73.7 (2018-03-30) SpiderCNN — 73.7 (2018-03-30) SpiderCNN — 73.7 (2018-03-30) GBNet — 80.5 (2019-11-28) GBNet — 80.5 (2019-11-28) GBNet — 80.5 (2019-11-28) DRNet — 80.3 (2020-05-14) DRNet — 80.3 (2020-05-14) DRNet — 80.3 (2020-05-14) MVTN — 82.8 (2020-11-26) MVTN — 82.8 (2020-11-26) MVTN — 82.8 (2020-11-26) SimpleView — 80.5 (2021-01-01) SimpleView — 80.5 (2021-01-01) SimpleView — 80.5 (2021-01-01) DynamicScale — 82.0 (2021-10-19) PatchAugment — 81.0 (2021-10-19) DynamicScale — 82.0 (2021-10-19) PatchAugment — 81.0 (2021-10-19) DynamicScale — 82.0 (2021-10-19) PatchAugment — 81.0 (2021-10-19) DeltaConv — 84.7 (2021-11-16) DeltaConv — 84.7 (2021-11-16) DeltaConv — 84.7 (2021-11-16) Point-BERT — 83.1 (2021-11-29) Point-BERT — 83.1 (2021-11-29) Point-BERT — 83.1 (2021-11-29) PRA-Net — 82.1 (2021-12-09) PRA-Net — 82.1 (2021-12-09) PRA-Net — 82.1 (2021-12-09) PointMLP — 85.7 (2022-02-15) PointMLP-elite — 83.8 (2022-02-15) PointMLP — 85.7 (2022-02-15) PointMLP-elite — 83.8 (2022-02-15) PointMLP — 85.7 (2022-02-15) PointMLP-elite — 83.8 (2022-02-15) Point-MAE — 85.2 (2022-03-13) Point-MAE — 85.2 (2022-03-13) Point-MAE — 85.2 (2022-03-13) Point-TnT — 83.5 (2022-04-08) Point-TnT — 83.5 (2022-04-08) Point-TnT — 83.5 (2022-04-08) APP-Net — 84.1 (2022-05-02) APP-Net — 84.1 (2022-05-02) APP-Net — 84.1 (2022-05-02) RepSurf-U (2x) — 86.0 (2022-05-11) RepSurf-U — 84.6 (2022-05-11) RepSurf-U (2x) — 86.0 (2022-05-11) RepSurf-U — 84.6 (2022-05-11) RepSurf-U (2x) — 86.0 (2022-05-11) RepSurf-U — 84.6 (2022-05-11) PointStack — 87.2 (2022-05-20) PointStack — 87.2 (2022-05-20) PointStack — 87.2 (2022-05-20) PointVector-S — 87.8 (2022-05-21) PointVector-S — 87.8 (2022-05-21) PointVector-S — 87.8 (2022-05-21) Point-M2AE — 86.43 (2022-05-28) Point-M2AE — 86.43 (2022-05-28) Point-M2AE — 86.43 (2022-05-28) PointNeXt — 88.2 (2022-06-09) PointNeXt — 88.2 (2022-06-09) PointNeXt — 88.2 (2022-06-09) P2P — 89.3 (2022-08-04) P2P — 89.3 (2022-08-04) P2P — 89.3 (2022-08-04) MVTN+SimpleView++ — 84.8 (2022-09-08) MVTN+SimpleView++ — 84.8 (2022-09-08) MVTN+SimpleView++ — 84.8 (2022-09-08) PointNeXt+HyCoRe — 88.3 (2022-09-21) PointNeXt+HyCoRe — 88.3 (2022-09-21) PointNeXt+HyCoRe — 88.3 (2022-09-21) PointCMT — 86.7 (2022-10-09) PointCMT — 86.7 (2022-10-09) PointCMT — 86.7 (2022-10-09) PointNet++ + SageMix — 83.7 (2022-10-13) DGCNN + SageMix — 83.6 (2022-10-13) PointNet++ + SageMix — 83.7 (2022-10-13) DGCNN + SageMix — 83.6 (2022-10-13) PointNet++ + SageMix — 83.7 (2022-10-13) DGCNN + SageMix — 83.6 (2022-10-13) PointNeXt+Local — 88.6 (2022-12-09) PointNeXt+Local — 88.6 (2022-12-09) PointNeXt+Local — 88.6 (2022-12-09) ULIP + PointNeXt — 89.7 (2022-12-10) ULIP + PointMLP — 89.4 (2022-12-10) ULIP + PointBERT — 86.4 (2022-12-10) ULIP + PointNeXt — 89.7 (2022-12-10) ULIP + PointMLP — 89.4 (2022-12-10) ULIP + PointBERT — 86.4 (2022-12-10) ULIP + PointNeXt — 89.7 (2022-12-10) ULIP + PointMLP — 89.4 (2022-12-10) ULIP + PointBERT — 86.4 (2022-12-10) I2P-MAE (no voting) — 90.11 (2022-12-13) I2P-MAE (no voting) — 90.11 (2022-12-13) I2P-MAE (no voting) — 90.11 (2022-12-13) ACT — 89.17 (2022-12-16) ACT (no voting) — 88.21 (2022-12-16) ACT — 89.17 (2022-12-16) ACT (no voting) — 88.21 (2022-12-16) ACT — 89.17 (2022-12-16) ACT (no voting) — 88.21 (2022-12-16) ReCon — 91.26 (2023-02-05) ReCon (no voting) — 90.63 (2023-02-05) ReCon — 91.26 (2023-02-05) ReCon (no voting) — 90.63 (2023-02-05) ReCon — 91.26 (2023-02-05) ReCon (no voting) — 90.63 (2023-02-05) PointConT — 90.3 (2023-03-08) PointConT (no voting) — 88.0 (2023-03-08) PointConT — 90.3 (2023-03-08) PointConT (no voting) — 88.0 (2023-03-08) PointConT — 90.3 (2023-03-08) PointConT (no voting) — 88.0 (2023-03-08) Point-PN — 87.1 (2023-03-14) Point-PN — 87.1 (2023-03-14) Point-PN — 87.1 (2023-03-14) SPoTr — 88.6 (2023-03-29) point2vec — 87.5 (2023-03-29) SPoTr — 88.6 (2023-03-29) point2vec — 87.5 (2023-03-29) SPoTr — 88.6 (2023-03-29) point2vec — 87.5 (2023-03-29) IDPT — 88.51 (2023-04-14) IDPT — 88.51 (2023-04-14) IDPT — 88.51 (2023-04-14) ULIP-2 + PointNeXt — 91.5 (2023-05-14) ULIP-2 + PointNeXt (no voting) — 90.8 (2023-05-14) ULIP-2 + Point-BERT — 89.0 (2023-05-14) ULIP-2 + PointNeXt — 91.5 (2023-05-14) ULIP-2 + PointNeXt (no voting) — 90.8 (2023-05-14) ULIP-2 + Point-BERT — 89.0 (2023-05-14) ULIP-2 + PointNeXt — 91.5 (2023-05-14) ULIP-2 + PointNeXt (no voting) — 90.8 (2023-05-14) ULIP-2 + Point-BERT — 89.0 (2023-05-14) PointGPT — 93.4 (2023-05-19) PointGPT — 93.4 (2023-05-19) PointGPT — 93.4 (2023-05-19) Point-LGMask — 85.3 (2023-06-08) Point-LGMask — 85.3 (2023-06-08) Point-LGMask — 85.3 (2023-06-08) PointMLP+TAP — 88.5 (2023-07-27) PointMLP+TAP — 88.5 (2023-07-27) PointMLP+TAP — 88.5 (2023-07-27) PointMLP∗ + JM3D — 89.5 (2023-08-06) PointMLP∗ + JM3D — 89.5 (2023-08-06) PointMLP∗ + JM3D — 89.5 (2023-08-06) DeLA — 90.4 (2023-08-31) DeLA — 90.4 (2023-08-31) DeLA — 90.4 (2023-08-31) Point-RAE (no voting) — 90.28 (2023-09-25) Point-RAE (no voting) — 90.28 (2023-09-25) Point-RAE (no voting) — 90.28 (2023-09-25) DualMLP — 86.4 (2023-10-10) DualMLP — 86.4 (2023-10-10) DualMLP — 86.4 (2023-10-10) OmniVec — 96.1 (2023-11-07) OmniVec — 96.1 (2023-11-07) OmniVec — 96.1 (2023-11-07) Point-FEMAE — 90.22 (2023-12-17) Point-FEMAE — 90.22 (2023-12-17) Point-FEMAE — 90.22 (2023-12-17) OmniVec2 — 97.2 (2024-01-01) OmniVec2 — 97.2 (2024-01-01) OmniVec2 — 97.2 (2024-01-01) PointMLS — 86.6 (2024-01-16) PointMLS — 86.6 (2024-01-16) PointMLS — 86.6 (2024-01-16) ReCon++ — 95.25 (2024-02-27) ReCon++ — 95.25 (2024-02-27) ReCon++ — 95.25 (2024-02-27) Mamba3D — 92.64 (2024-04-23) Mamba3D (no voting) — 91.81 (2024-04-23) Mamba3D — 92.64 (2024-04-23) Mamba3D (no voting) — 91.81 (2024-04-23) Mamba3D — 92.64 (2024-04-23) Mamba3D (no voting) — 91.81 (2024-04-23) Point-JEPA — 86.6 (2024-04-25) Point-JEPA — 86.6 (2024-04-25) Point-JEPA — 86.6 (2024-04-25) KPConvX-L — 89.3 (2024-05-21) KPConvX-L — 89.3 (2024-05-21) KPConvX-L — 89.3 (2024-05-21) GPSFormer — 95.4 (2024-07-18) GPSFormer-elite — 93.3 (2024-07-18) GPSFormer — 95.4 (2024-07-18) GPSFormer-elite — 93.3 (2024-07-18) GPSFormer — 95.4 (2024-07-18) GPSFormer-elite — 93.3 (2024-07-18) PCP-MAE — 90.35 (2024-08-16) PCP-MAE — 90.35 (2024-08-16) PCP-MAE — 90.35 (2024-08-16) ReCon+PPT — 89.52 (2024-08-21) ReCon+PPT — 89.52 (2024-08-21) ReCon+PPT — 89.52 (2024-08-21) 3D-JEPA — 89.52 (2024-09-24) 3D-JEPA — 89.52 (2024-09-24) 3D-JEPA — 89.52 (2024-09-24) PointGST — 96.18 (2024-10-10) PointGST — 96.18 (2024-10-10) PointGST — 96.18 (2024-10-10) Ours — 89.0 (2024-11-01) Ours — 89.0 (2024-11-01) Ours — 89.0 (2024-11-01) OTMae3D — 89.0 (2024-12-26) OTMae3D — 89.0 (2024-12-26) OTMae3D — 89.0 (2024-12-26) AsymDSD-B* (no voting) — 93.72 (2025-06-26) AsymDSD-S (no voting) — 90.53 (2025-06-26) AsymDSD-B* (no voting) — 93.72 (2025-06-26) AsymDSD-S (no voting) — 90.53 (2025-06-26) AsymDSD-B* (no voting) — 93.72 (2025-06-26) AsymDSD-S (no voting) — 90.53 (2025-06-26) PointNeXt+GAM — 88.4 (2015-12-14) P2P — 89.3 (2022-08-04) ULIP + PointNeXt — 89.7 (2022-12-10) I2P-MAE (no voting) — 90.11 (2022-12-13) ReCon — 91.26 (2023-02-05) ULIP-2 + PointNeXt — 91.5 (2023-05-14) PointGPT — 93.4 (2023-05-19) OmniVec — 96.1 (2023-11-07) OmniVec2 — 97.2 (2024-01-01)
RankModel Overall AccuracyMean AccuracyOBJ-BG (OA)OBJ-ONLY (OA)FLOPsNumber of params Extra Training Data PaperCodeYear
1 OmniVec2 97.2––––– ✓ OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning 2024
2 PointGST 96.18–99.4897.76–– ✓ Parameter-Efficient Fine-Tuning in Spectral Domain for Point Cloud Learning jerryfeng2003/pointgst 2024
3 OmniVec 96.1––––– ✓ OmniVec: Learning robust representations with cross modal sharing 2023
4 GPSFormer 95.493.8––0.7G2.36M GPSFormer: A Global Perception and Local Structure Fitting-based Transformer for Point Cloud Understanding changshuowang/GPSFormer 2024
5 ReCon++ 95.25–98.8097.59–– ✓ ShapeLLM: Universal 3D Object Understanding for Embodied Interaction qizekun/ShapeLLM · qizekun/ReCon · runpeidong/act 2024
6 AsymDSD-B* (no voting) 93.72–96.7394.32–– ✓ Asymmetric Dual Self-Distillation for 3D Self-Supervised Representation Learning RFLeijenaar/AsymDSD 2025
7 PointGPT 93.4–97.296.6–– ✓ PointGPT: Auto-regressively Generative Pre-training from Point Clouds CGuangyan-BIT/PointGPT 2023
8 GPSFormer-elite 93.3092.51–––0.68M GPSFormer: A Global Perception and Local Structure Fitting-based Transformer for Point Cloud Understanding changshuowang/GPSFormer 2024
9 Mamba3D 92.64–94.4992.433.9G16.9M Mamba3D: Enhancing Local Features for 3D Point Cloud Analysis via State Space Model xhanxu/Mamba3D 2024
10 Mamba3D (no voting) 91.81–92.9492.083.9G16.9M Mamba3D: Enhancing Local Features for 3D Point Cloud Analysis via State Space Model xhanxu/Mamba3D 2024
11 ULIP-2 + PointNeXt 91.591.2–––1.4M ✓ ULIP-2: Towards Scalable Multimodal Pre-training for 3D Understanding salesforce/ulip 2023
12 ReCon 91.26–95.3593.80–– ✓ Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining qizekun/ReCon · runpeidong/act · aHapBean/PCP-MAE · +2 2023
13 ULIP-2 + PointNeXt (no voting) 90.890.3–––1.4M ✓ ULIP-2: Towards Scalable Multimodal Pre-training for 3D Understanding salesforce/ulip 2023
14 ReCon (no voting) 90.63–95.1893.29–– ✓ Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining qizekun/ReCon · runpeidong/act · aHapBean/PCP-MAE · +2 2023
15 AsymDSD-S (no voting) 90.53–94.3291.91–– ✓ Asymmetric Dual Self-Distillation for 3D Self-Supervised Representation Learning RFLeijenaar/AsymDSD 2025
16 DeLA 90.489.3––1.5G5.3M Decoupled Local Aggregation for Point Cloud Learning matrix-asc/dela 2023
17 PCP-MAE 90.35–95.5294.32–– PCP-MAE: Learning to Predict Centers for Point Masked Autoencoders aHapBean/PCP-MAE 2024
18 PointConT 90.388.5–––– Point Cloud Classification Using Content-based Transformer via Clustering in Feature Space yahuiliu99/pointcont 2023
19 Point-RAE (no voting) 90.28–95.5393.63–– ✓ Regress Before Construct: Regress Autoencoder for Point Cloud Self-supervised Learning liuyyy111/point-rae 2023
20 Point-FEMAE 90.22–95.1893.29–– ✓ Towards Compact 3D Representations via Point Feature Enhancement Masked Autoencoders zyh16143998882/aaai24-pointfemae 2023
1–20 / 231 다음 → 페이지당 10 20 50 100