Video Depth Estimation on Bonn (AbsRel, δ < 1.25)
4.5AbsRelpi^3
Evaluation results
| Method | Links | ||
|---|---|---|---|
| pi^3Params=959M, Inference Speed Complexity=O(N^2), Alignment Protocol=joint scale-and-shift2026.03 | 4.5 | 97.6 | |
| VDAZero-shot=true, Train Frames=60M2026.03 | 4.7 | 97 | |
| DVDZero-shot=true, Train Frames=367K2026.03 | 4.7 | 97.1 | |
| VGGTParams=1.26B, Inference Speed Complexity=O(N^2), Alignment Protocol=joint scale-and-shift2026.03 | 4.9 | 97.4 | |
| pi^3Params=959M2025.07 | 4.9 | 97.5 | |
| DA3Type=FA2026.06 | 4.9 | 97.2 | |
| DA3 + Ours (LMM)Type=FA2026.06 | 4.9 | 97.4 | |
| FFNTest-Time Training (TTT)=true2026.06 | 5.1 | 98.2 | |
| FrameVGGTM=242026.03 | 5.12 | 97.99 | |
| FrameVGGTM=202026.03 | 5.14 | 97.99 | |
| ZipMapParams=1.40B, Inference Speed Complexity=O(N), Alignment Protocol=joint scale-and-shift2026.03 | 5.2 | 97.3 | |
| FrameVGGTM=162026.03 | 5.25 | 98.01 | |
| FrameVGGTM=122026.03 | 5.26 | 97.93 | |
| PAGE-4DParams=1.26B, Align=Monocular Depth2025.10 | 5.3 | 97 | |
| TurboVGGT2026.05 | 5.3 | 97.5 | |
| MoE3DType=FA2026.06 | 5.3 | 97 | |
| DA3 + Ours (GMM)Type=FA2026.06 | 5.3 | 97.2 | |
| TTT3RParams=793M, Inference Speed Complexity=O(N), Alignment Protocol=joint scale-and-shift2026.03 | 5.4 | 97.3 | |
| VGGTtype=FA2026.03 | 5.5 | 97 | |
| MoRetype=FA2026.03 | 5.5 | 97.1 | |
| StreamVGGTSequence Length=1002026.03 | 5.5 | 97.4 | |
| OVGGTSequence Length=1002026.03 | 5.5 | 97.4 | |
| VGGTAlignment=Per-sequence scale, Online=false, Venue=CVPR’252026.03 | 5.5 | 97.1 | |
| VGGTType=FA2026.06 | 5.5 | 97.1 | |
| InfiniteVGGTSequence Length=1002026.03 | 5.6 | 97.5 | |
| InfiniteVGGT2026.03 | 5.6 | 98.02 | |
| VGGTParams=1.26B2025.07 | 5.7 | 96.6 | |
| VGGT2026.03 | 5.7 | 96.8 | |
| VGGTType=Dense-view2025.07 | 5.7 | 96.8 | |
| SparseVGGT2026.05 | 5.7 | 96.8 | |
| StreamVGGTtype=Streaming2026.03 | 5.8 | 97.2 | |
| CUT3RParams=793M, Align=Monocular Depth2025.10 | 5.8 | 96.7 | |
| FastVGGT2026.05 | 5.8 | 96.9 | |
| RollingDepthZero-shot=true, Train Frames=-2026.03 | 5.9 | 96.6 | |
| DepthCrafterZero-shot=true, Train Frames=∼ 30M2026.03 | 5.9 | 95.9 | |
| Easi3RAlignment=Per-sequence scale, Online=false, Venue=ICCV’252026.03 | 5.9 | 97 | |
| StreamVGGTAlignment=Per-sequence scale, Online=true, Venue=ArXiv’252026.03 | 5.9 | 97.2 | |
| StreamVGGTType=Streaming2025.07 | 5.9 | 97.2 | |
| VGGT2026.05 | 5.9 | 96.7 | |
| ZipMap2026.05 | 5.9 | 97.3 | |
| Point3RAlignment=Per-sequence scale, Online=true, Venue=NeurIPS’252026.03 | 6 | 96 | |
| Point3RType=Streaming2025.07 | 6 | 96 | |
| TTT3RStreaming=true, Alignment=scale-only2026.03 | 6.1 | 97 | |
| SSRAlignment=Per-sequence scale, Online=true, Venue=Ours2026.03 | 6.1 | 96.8 | |
| TTT3R2026.05 | 6.1 | 96.9 | |
| Evict3R†Sequence Length=1002026.03 | 6.3 | 96.3 | |
| VGG-T32026.05 | 6.3 | 96.3 | |
| Online TTT-MAETest-Time Training (TTT)=true, Sliding window (s.w.)=true2026.06 | 6.4 | 97 | |
| Point3RParam.=600M2026.03 | 6.6 | 95.8 | |
| Complet4R2026.03 | 6.6 | 95.9 | |
| CUT3RParams=793M, Inference Speed Complexity=O(N), Alignment Protocol=joint scale-and-shift2026.03 | 6.7 | 96.1 | |
| ZipMapStreaming=true, Alignment=scale-only2026.03 | 6.7 | 96.5 | |
| OVGGTSequence Length=5002026.03 | 6.7 | 95.9 | |
| MonST3R2025.08 | 6.7 | 96.3 | |
| MonSTR3RAlignment=Per-sequence scale, Online=false, Venue=ICLR’252026.03 | 6.7 | 96.3 | |
| MonST3R-GAType=Pair-wise2025.07 | 6.7 | 96.3 | |
| MoRetype=Streaming2026.03 | 6.8 | 96.1 | |
| TTT3RAlignment=Per-sequence scale, Online=true, Venue=ICLR’262026.03 | 6.8 | 95.4 | |
| SSRAlignment=Metric scale, Online=true, Venue=Ours2026.03 | 6.8 | 96.5 | |
| DAv2-LZero-shot=true, Train Frames=-2026.03 | 6.9 | 95.7 | |
| MonST3RParams=571M, Align=Monocular Depth2025.10 | 6.9 | 95.4 | |
| Stream3Rtype=Streaming2026.03 | 7 | 95.2 | |
| InfiniteVGGTSequence Length=5002026.03 | 7 | 96 | |
| Video Depth-AnythingZero-shot=true2026.06 | 7 | 96.1 | |
| Wint3Rtype=Streaming2026.03 | 7.1 | 91.2 | |
| OVGGTSequence Length=3002026.03 | 7.1 | 95.6 | |
| VGGTParams=1.26B, Align=Monocular Depth2025.10 | 7.1 | 94.7 | |
| CUT3RStreaming=true, Alignment=scale-only2026.03 | 7.2 | 95.1 | |
| Evict3R†Sequence Length=3002026.03 | 7.2 | 95.1 | |
| Evict3R†Sequence Length=5002026.03 | 7.2 | 95.7 | |
| MonST3RParams=571M2025.07 | 7.2 | 95.7 | |
| OmniStreamParam.=400M2026.03 | 7.2 | 96.7 | |
| OmniStream2026.03 | 7.2 | — | |
| InfiniteVGGTSequence Length=3002026.03 | 7.3 | 95.7 | |
| CUT3RParams=793M, Align=scale&shift, Video Depth2025.10 | 7.5 | 94.3 | |
| TTT3R2026.03 | 7.52 | 95.87 | |
| CUT3Rtype=Streaming2026.03 | 7.6 | 94.9 | |
| VGGT + Ours (GMM)Type=FA2026.06 | 7.6 | 94.2 | |
| CUT3RParams=793M, Align=scale, Video Depth2025.10 | 7.7 | 93.7 | |
| CUT3RParams=793M2025.07 | 7.8 | 93.7 | |
| Cut3RParam.=600M2026.03 | 7.8 | 93.7 | |
| CUT3R2026.03 | 7.8 | — | |
| CUT3RAlignment=Per-sequence scale, Online=true, Venue=CVPR’252026.03 | 7.8 | 93.7 | |
| CUT3R2026.03 | 7.8 | 93.7 | |
| CUT3RType=Streaming2025.07 | 7.8 | 93.7 | |
| Point3R2026.03 | 8.09 | 95.13 | |
| CUT3R2026.03 | 8.31 | 94.02 | |
| UniUGG2025.08 | 8.6 | 91.4 | |
| TTT3RAlignment=Metric scale, Online=true, Venue=ICLR’262026.03 | 9 | 94.2 | |
| PAGE-4DParams=1.26B, Align=scale&shift, Video Depth2025.10 | 9 | 90.3 | |
| PAGE-4DParams=1.26B, Align=scale, Video Depth2025.10 | 9.2 | 90.4 | |
| Marigold v1.1Zero-shot=true, Train Frames=-2026.03 | 9.5 | 93.6 | |
| VGGTParams=1.26B, Align=scale&shift, Video Depth2025.10 | 10.2 | 89 | |
| CUT3RAlignment=Metric scale, Online=true, Venue=CVPR’252026.03 | 10.3 | 88.5 | |
| VGGTParams=1.26B, Align=scale, Video Depth2025.10 | 10.7 | 88.3 | |
| Flaretype=FA2026.03 | 11.6 | 89.7 | |
| MASt3RParams=689M, Align=Monocular Depth2025.10 | 12.3 | 83.3 | |
| MapAnythingtype=FA2026.03 | 12.5 | 85.9 | |
| DAv2-LZero-shot=true2026.06 | 12.7 | 86.4 | |
| FLAREParams=1.40B, Align=Monocular Depth2025.10 | 13 | 83.6 |