Base-to-new class generalization on ScanObjectNN S-OBJ_ONLY (test)
84.05Base AccuracyReFine3D
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ReFine3DPre-trained model=ULIP-2, Evaluation protocol=Fine-tuning2026.06 | 84.05 | 82.97 | 83.51 | |
| ULIP-2 + RCprompt learning=true, regulation constraint framework=true2024.10 | 83.6 | 81.1 | 82.33 | |
| PointPRCPre-trained model=ULIP-2, Evaluation protocol=Prompt tuning2026.06 | 83.6 | 81.1 | 82.33 | |
| ReFine3DPre-trained model=ULIP, Evaluation protocol=Fine-tuning2026.06 | 82.65 | 68.26 | 74.77 | |
| ULIP + RCprompt learning=true, regulation constraint framework=true2024.10 | 79.23 | 65.93 | 71.97 | |
| PointPRCPre-trained model=ULIP, Evaluation protocol=Prompt tuning2026.06 | 79.23 | 65.93 | 71.97 | |
| ULIP-2prompt learning=true2024.10 | 78.6 | 76.27 | 77.42 | |
| ULIP-2Pre-trained model=ULIP-2, Evaluation protocol=Prompt tuning2026.06 | 78.6 | 76.27 | 77.42 | |
| P-CLIPprompt learning=true2024.10 | 76.23 | 20.23 | 31.97 | |
| P-CLIPPre-trained model=ULIP, Evaluation protocol=Prompt tuning2026.06 | 76.23 | 20.23 | 31.97 | |
| ULIPprompt learning=true2024.10 | 74.13 | 50.8 | 60.29 | |
| ULIPPre-trained model=ULIP, Evaluation protocol=Prompt tuning2026.06 | 74.13 | 50.8 | 60.29 | |
| P-CLIP2prompt learning=true2024.10 | 71.4 | 44.39 | 54.74 | |
| P-CLIP2Pre-trained model=ULIP, Evaluation protocol=Prompt tuning2026.06 | 71.4 | 44.39 | 54.74 |