DL3DV
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Novel View Synthesis | DL3DV | PSNR28.141 | 61 | |
| Novel View Synthesis | DL3DV (test) | PSNR30.39 | 54 | |
| Novel View Synthesis | DL3DV (evaluation) | PSNR30.75 | 22 | |
| 3D Reconstruction | DL3DV-140 | PSNR27.07 | 18 | |
| Novel View Synthesis | DL3DV v1 (hold-out scenes) | PSNR28.47 | 16 | |
| Novel-view synthesis | DL3DV-140 at 960x540 resolution (test) | PSNR28.3 | 13 | |
| Single-view Novel View Synthesis | DL3DV (Long-term (200th frame)) | PSNR14.53 | 13 | |
| Single-view Novel View Synthesis | DL3DV Short-term (50th frame) | PSNR18.1 | 13 | |
| Text Reconstruction | DL3DV-10K | CER12.3 | 12 | |
| I2V Camera Control | DL3DV (test) | RRE0.0886 | 10 | |
| 3D Reconstruction | DL3DV 9 views | PSNR19.6 | 9 | |
| 3D Reconstruction | DL3DV 6 views | PSNR18.74 | 9 | |
| 3D Reconstruction | DL3DV 3 views | PSNR16.87 | 9 | |
| Novel View Synthesis | DL3DV (in-domain) | PSNR25.75 | 8 | |
| Novel View Synthesis | DL3DV 140 (test) | PSNR26.89 | 6 | |
| Novel View Synthesis | DL3DV-10K (30K iterations) | Training Time (min)11.4 | 6 | |
| Novel View Synthesis | DL3DV-10K 7K iterations | Training Time (min)2.1 | 6 | |
| 3D Scene Reconstruction | DL3DV-10K | PSNR30.47 | 6 | |
| Pose Estimation | DL3DV | RPA @ 5 deg72 | 6 | |
| Novel View Synthesis | DL3DV high-resolution | PSNR24.411 | 6 | |
| Novel View Synthesis | DL3DV-BLUR proposed | PSNR (3-view)17.48 | 5 | |
| Image-to-Video Generation | DL3DV-10K | PSNR14.9 | 5 | |
| Multi-view Promptable Segmentation | DL3DV | mIoU0.751 | 5 | |
| Novel View Synthesis | DL3DV 6view | PSNR17.98 | 4 | |
| Novel View Synthesis | DL3DV 3view | PSNR16.37 | 4 |
Showing 25 of 39 rows