Image-Based Shape Retrieval on StanfordCars 21 (test)
11.4Top-1 Instance AccuracyPoint-BERT(L)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Point-BERT(L)Framework=OpenShape, Pre-train Data=Ensembled, Zero-shot=true2026.03 | 11.4 | — | 49.1 | — | 21.8 | — | |
| Point-BERT(S)Framework=OpenShape, Pre-train Data=Shapenet, Zero-shot=true2026.03 | 11.2 | — | 49.9 | — | 21.9 | — | |
| SparseConvFramework=OpenShape, Pre-train Data=Ensembled, Zero-shot=true2026.03 | 5.6 | — | 33.1 | — | 12.2 | — | |
| SparseConvFramework=OpenShape, Pre-train Data=Shapenet, Zero-shot=true2026.03 | 4 | — | 22.1 | — | 8.4 | — | |
| Point-BERTFramework=ULIP, Pre-train Data=Ensembled, Zero-shot=true2026.03 | 1.8 | — | 8.6 | — | 3.3 | — | |
| Point-BERTFramework=ULIP, Pre-train Data=Shapenet, Zero-shot=true2026.03 | 1.7 | — | 8 | — | 3 | — | |
| Point-BERTFramework=ULIP2, Pre-train Data=Ensembled, Zero-shot=true2026.03 | 1.7 | — | 8.4 | — | 3.4 | — | |
| PointNeXtFramework=ULIP2, Pre-train Data=Ensembled, Zero-shot=true2026.03 | 0.9 | — | 12.1 | — | 3.2 | — |