Zero-Shot Retrieval on ModelNet40 46 (test)
30.5Top-1 Accuracy (Instance)Point-BERT(L)
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Point-BERT(L)Framework=OpenShape, Pre-train Data=Ensembled2026.03 | 30.5 | 97.2 | 67.1 | 92.9 | 41.6 | 80.1 | |
| Point-BERT(S)Framework=OpenShape, Pre-train Data=Shapenet2026.03 | 18.7 | 71.1 | 55.2 | 90.7 | 29.1 | 73.9 | |
| SparseConvFramework=OpenShape, Pre-train Data=Ensembled2026.03 | 17.9 | 18 | 51 | 52.6 | 27.6 | 27.7 | |
| SparseConvFramework=OpenShape, Pre-train Data=Shapenet2026.03 | 9.5 | 63.8 | 38 | 88.2 | 17.1 | 67.4 | |
| PointNeXtFramework=ULIP2, Pre-train Data=Ensembled2026.03 | 3.5 | 3.6 | 14.6 | 16 | 6.2 | 6.6 | |
| Point-BERTFramework=ULIP2, Pre-train Data=Ensembled2026.03 | 3 | 3.1 | 9 | 10.2 | 4.6 | 4.9 | |
| Point-BERTFramework=ULIP, Pre-train Data=Ensembled2026.03 | 1.6 | 22.8 | 6.4 | 61.1 | 2.9 | 30.3 | |
| Point-BERTFramework=ULIP, Pre-train Data=Shapenet2026.03 | 0.3 | 9.5 | 2.1 | 35.9 | 0.7 | 15.6 |