3D Object Classification on ScanObjectNN PB
89.7AccuracyRECON
Evaluation results
| Method | Links | |
|---|---|---|
| RECON#LP (M)=43.6, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 89.7 | |
| P3T#LP (M)=2.0, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=true2026.04 | 89.6 | |
| PPT#LP (M)=1.8, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=true2026.04 | 89.1 | |
| ACT#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 88.2 | |
| P3T#LP (M)=2.0, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=false2026.04 | 88.1 | |
| PPT#LP (M)=1.8, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=false2026.04 | 87.2 | |
| PointGPT#LP (M)=29.2, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 86.9 | |
| Point-M2AE#LP (M)=15.3, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 86.4 | |
| PointMLP#LP (M)=12.6, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 85.4 | |
| Point-MAE#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 85.2 | |
| ULIP-2#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 85 | |
| ULIP#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 84.7 | |
| MaskPoint#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 84.6 | |
| Point-BERT#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 83.1 | |
| MVTN#LP (M)=11.2, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 82.8 | |
| DAPT#LP (M)=0.8, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=false2026.04 | 82.3 | |
| Point-PRC#LP (M)=0.02, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=true2026.04 | 81.9 | |
| IDPT#LP (M)=1.3, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=false2026.04 | 80.3 | |
| Point-PRC#LP (M)=0.02, Training Protocol=Self-Supervised Pre-training + Parameter-efficient fine-tuning, PPT Setting=false2026.04 | 79.5 | |
| OcCo#LP (M)=22.1, Training Protocol=Self-Supervised Pre-training + Full fine-tuning, PPT Setting=false2026.04 | 78.8 | |
| PointCNN#LP (M)=0.6, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 78.5 | |
| DGCNN#LP (M)=1.8, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 78.1 | |
| PointNet++#LP (M)=1.5, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 77.9 | |
| PointNet#LP (M)=3.5, Training Protocol=Supervised Learning, PPT Setting=false2026.04 | 68 | |
| P3TCross-dataset generalization=true, Source dataset=Objaverse-LVIS2026.04 | 37 | |
| Point-PRCCross-dataset generalization=true, Source dataset=Objaverse-LVIS2026.04 | 30.8 | |
| PPTCross-dataset generalization=true, Source dataset=Objaverse-LVIS2026.04 | 26.2 | |
| ULIP-2 (ZS)Zero-shot setting=true, Cross-dataset generalization=true, Source dataset=Objaverse-LVIS2026.04 | 25.3 | |
| ULIP-2Cross-dataset generalization=true, Source dataset=Objaverse-LVIS2026.04 | 16.6 |