| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| 3D Object Classification | Objaverse-LVIS (test) | Top-1 Accuracy83.1 | 95 | |
| 3D Object Captioning | Objaverse | Sentence-BERT Performance Score51.91 | 33 | |
| 3D Object Classification | Objaverse | Average Accuracy71 | 30 | |
| 3D Captioning | Objaverse (test) | S-BERT Score100 | 28 | |
| Object Recognition | Objaverse-LVIS | Top-1 Acc53.7 | 25 | |
| 3D Object Classification | Objaverse LVIS 10 (test) | Top-1 Acc50.7 | 19 | |
| 3D Classification | Objaverse LVIS | Top-1 Acc59.5 | 19 | |
| 3D Mesh Generation | Objaverse | Chamfer Distance0.009 | 18 | |
| 3D Object Recognition | Objaverse-LVIS | Accuracy55.42 | 16 | |
| Tokenizer reconstruction | Objaverse | CD (x10^-2)0.034 | 15 | |
| 3D-Text Retrieval | Objaverse-LVIS 1.0 (test) | CLIP Top-5 Retrieval85.6 | 15 | |
| 3D-Text Matching | Objaverse-LVIS 1.0 (test) | CLIP Matching Accuracy67.6 | 15 | |
| Text-guided visual synthesis | Objaverse | FID36.07 | 14 | |
| Novel view synthesis | Objaverse (test) | PSNR27.59 | 14 | |
| 4D Mesh Reconstruction | Objaverse (test) | CD0.0786 | 13 | |
| 3D Tokenization | Objaverse 128 watertight assets filtered via Step1X-3D | CD12.3 | 12 | |
| 3D Object Recognition | Objaverse | GPT-4 Score58.48 | 12 | |
| Novel View Synthesis | Objaverse | PSNR19.88 | 12 | |
| Text-to-3D Generation | Objaverse | CLIP Score30.56 | 12 | |
| Geometric reconstruction | Objaverse PBR | Chamfer Distance82.49 | 11 | |
| Text-to-3D | Objaverse 1.0 (test) | CLIP Score81.6 | 11 | |
| Pose Optimization | Objaverse | D5 Error0 | 10 | |
| Image-conditioned 3D Generation | Objaverse (test) | FID19.93 | 10 | |
| Relighting | Objaverse | LPIPS0.0479 | 9 | |
| Active object reconstruction | Objaverse views (test) | PSNR (Avg)27 | 9 |