Visual Localization on Cambridge Landmarks (test)
0.136Avg Median Positional Error (m)VS-Net
Evaluation results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VS-NetMethod Category=SSL2021.05 | 0.136 | 0.22 | 0.1 | 0.16 | 0.2 | 0.16 | 0.3 | 0.06 | 0.3 | 0.08 | 0.3 | 0.24 | — | — | — | — | — | — | |
| HSC-NetMethod Category=Scene Coordinate2021.05 | 0.16 | 0.28 | 0.2 | 0.18 | 0.3 | 0.19 | 0.3 | 0.06 | 0.3 | 0.09 | 0.3 | 0.28 | — | — | — | — | — | — | |
| DSAC++Method Category=Scene Coordinate2021.05 | 0.194 | 0.4 | 0.2 | 0.18 | 0.3 | 0.2 | 0.3 | 0.06 | 0.3 | 0.13 | 0.4 | 0.3 | — | — | — | — | — | — | |
| PMNetMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.31 | — | — | 0.31 | 0.55 | 0.44 | 0.79 | 0.17 | 0.86 | 0.31 | 0.96 | 0.79 | — | — | — | — | — | — | |
| DFNet+NeFeSMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.35 | — | — | 0.37 | 0.54 | 0.52 | 0.88 | 0.15 | 0.53 | 0.37 | 1.14 | 0.77 | — | — | — | — | — | — | |
| HF-NetMethod Category=SfM2021.05 | 0.356 | 0.76 | 0.3 | 0.34 | 0.4 | 0.43 | 0.6 | 0.09 | 0.4 | 0.16 | 0.5 | 0.31 | — | — | — | — | — | — | |
| RegMethod Category=Scene Coordinate2021.05 | 0.378 | 1.25 | 0.6 | 0.21 | 0.3 | 0.21 | 0.3 | 0.06 | 0.3 | 0.16 | 0.5 | 0.4 | — | — | — | — | — | — | |
| Reloc3r-512Method Category=RPR, Training Setting=Unseen, Input Resolution=512, Dataset-specific training time=None2024.12 | 0.38 | 1.22 | 0.73 | 0.42 | 0.36 | 0.62 | 0.55 | 0.13 | 0.58 | 0.34 | 0.58 | 0.52 | — | — | — | — | — | — | |
| LENSMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.39 | — | — | 0.33 | 0.5 | 0.44 | 0.9 | 0.27 | 1.6 | 0.53 | 1.6 | 1.2 | — | — | — | — | — | — | |
| Reloc3r-224Method Category=RPR, Training Setting=Unseen, Input Resolution=224, Dataset-specific training time=None2024.12 | 0.48 | 1.71 | 0.94 | 0.47 | 0.41 | 0.87 | 0.66 | 0.18 | 0.53 | 0.41 | 0.73 | 0.58 | — | — | — | — | — | — | |
| ImageNet+NCMMethod Category=RPR, Training Setting=Unseen, Hybrid Pose Estimation=true, Dataset-specific training time=None2024.12 | 0.83 | — | — | — | — | — | — | — | — | — | — | 1.36 | — | — | — | — | — | — | |
| AnchorNetMethod Category=RPR, Training Setting=Seen, Dataset-specific training time=Hours2024.12 | 0.84 | — | — | 0.57 | 0.88 | 1.21 | 2.55 | 0.52 | 2.27 | 1.04 | 2.69 | 2.1 | — | — | — | — | — | — | |
| NC-EssNet (CL)Method Category=RPR, Training Setting=Seen, Dataset-specific training time=Hours2024.12 | 0.85 | — | — | — | — | — | — | — | — | — | — | 2.82 | — | — | — | — | — | — | |
| Relpose-GNNMethod Category=RPR, Training Setting=Seen, Dataset-specific training time=Hours2024.12 | 0.91 | 3.2 | 2.2 | 0.48 | 1 | 1.14 | 2.5 | 0.48 | 2.5 | 1.52 | 3.2 | 2.3 | — | — | — | — | — | — | |
| EssNet (CL)Method Category=RPR, Training Setting=Seen, Dataset-specific training time=Hours2024.12 | 1.08 | — | — | — | — | — | — | — | — | — | — | 3.41 | — | — | — | — | — | — | |
| +OursMode=Multi-scene APR, Features=QKA loss, fixed 2D sinusoidal positional encoding2024.11 | 1.19 | — | — | 0.88 | 1.29 | 1.55 | 1.87 | 0.79 | 2.51 | 1.57 | 3.5 | 2.29 | — | — | — | — | — | — | |
| MS-TransformerAPR category=Multi-scene APR2021.03 | 1.28 | — | — | — | — | — | — | — | — | — | — | 2.73 | 1 | 1 | — | — | — | — | |
| MSTMode=Multi-scene APR2024.11 | 1.28 | — | — | 0.83 | 1.47 | 1.81 | 2.39 | 0.86 | 3.07 | 1.62 | 3.99 | 2.73 | — | — | — | — | — | — | |
| LSTM-PNAPR category=Single-scene APR2021.03 | 1.3 | — | — | — | — | — | — | — | — | — | — | 5.52 | 2 | 9 | — | — | — | — | |
| SVS-PoseAPR category=Single-scene APR2021.03 | 1.33 | — | — | — | — | — | — | — | — | — | — | 5.17 | 3 | 7 | — | — | — | — | |
| IRPNetAPR category=Single-scene APR2021.03 | 1.42 | — | — | — | — | — | — | — | — | — | — | 3.45 | 4 | 4 | — | — | — | — | |
| PoseNet-LearnableAPR category=Single-scene APR2021.03 | 1.43 | — | — | — | — | — | — | — | — | — | — | 2.85 | 5 | 2 | — | — | — | — | |
| GeoPoseNetAPR category=Single-scene APR2021.03 | 1.63 | — | — | — | — | — | — | — | — | — | — | 2.86 | 6 | 3 | — | — | — | — | |
| MapNetAPR category=Single-scene APR2021.03 | 1.63 | — | — | — | — | — | — | — | — | — | — | 3.64 | 6 | 5 | — | — | — | — | |
| BayesianPNAPR category=Single-scene APR2021.03 | 1.92 | — | — | — | — | — | — | — | — | — | — | 6.28 | 8 | 10 | — | — | — | — | |
| GPoseNetAPR category=Single-scene APR2021.03 | 2.08 | — | — | — | — | — | — | — | — | — | — | 4.59 | 6 | 3 | — | — | — | — | |
| PoseNetAPR category=Single-scene APR2021.03 | 2.09 | — | — | — | — | — | — | — | — | — | — | 6.84 | 10 | 11 | — | — | — | — | |
| ExReNet (SUNCG)Method Category=RPR, Training Setting=Unseen, Backbone / Pre-training=SUNCG, Dataset-specific training time=None2024.12 | 2.22 | 9.79 | 4.46 | 2.33 | 2.48 | 3.54 | 3.49 | 0.72 | 2.41 | 2.3 | 3.72 | 3.03 | — | — | — | — | — | — | |
| ExReNet (SN)Method Category=RPR, Training Setting=Unseen, Backbone / Pre-training=SN, Dataset-specific training time=None2024.12 | 2.36 | 10.97 | 6.52 | 2.48 | 2.92 | 3.47 | 3.9 | 0.9 | 3.27 | 2.6 | 4.98 | 3.77 | — | — | — | — | — | — | |
| MSPNAPR category=Multi-scene APR2021.03 | 2.47 | — | — | — | — | — | — | — | — | — | — | 5.34 | 11 | 8 | — | — | — | — | |
| MSPNMode=Multi-scene APR2024.11 | 2.47 | — | — | 1.73 | 3.65 | 2.55 | 4.05 | 2.92 | 7.49 | 2.67 | 6.18 | 5.34 | — | — | — | — | — | — | |
| Map-free (Regress)Method Category=RPR, Training Setting=Unseen, Backbone / Pre-training=Regress, Dataset-specific training time=None2024.12 | 2.51 | 8.4 | 4.56 | 2.44 | 2.54 | 3.73 | 5.23 | 0.97 | 3.17 | 2.91 | 5.1 | 4.01 | — | — | — | — | — | — | |
| Map-free (Match)Method Category=RPR, Training Setting=Unseen, Hybrid Pose Estimation=true, Backbone / Pre-training=Match, Dataset-specific training time=None2024.12 | 2.61 | 9.09 | 5.33 | 2.51 | 3.11 | 3.89 | 6.44 | 1.04 | 3.61 | 3 | 6.14 | 4.83 | — | — | — | — | — | — | |
| NC-EssNet (7S)Method Category=RPR, Training Setting=Unseen, Dataset-specific training time=None2024.12 | 7.98 | — | — | — | — | — | — | — | — | — | — | 24.35 | — | — | — | — | — | — | |
| EssNet (7S)Method Category=RPR, Training Setting=Unseen, Dataset-specific training time=None2024.12 | 10.36 | — | — | — | — | — | — | — | — | — | — | 85.75 | — | — | — | — | — | — | |
| ASMethod Category=SfM2021.05 | — | — | — | 0.42 | 0.55 | 0.44 | 1.01 | 0.12 | 0.4 | 0.19 | 0.54 | — | — | — | — | — | — | — | |
| ASNo Desc. Maint.=false, Privacy=false, Method Category=Descriptor-based (DB), MB used=8132023.06 | — | — | — | — | 0.22 | — | 0.36 | — | 0.21 | — | 0.25 | — | — | — | 0.13 | 0.2 | 0.04 | 0.08 | |
| DGC-GNNNo Desc. Maint.=true, Privacy=true, Method Category=Descriptor-free (DF), MB used=692023.06 | — | — | — | — | 0.47 | — | 2.83 | — | 1.57 | — | 4.03 | — | — | — | 0.18 | 0.75 | 0.15 | 1.06 | |
| DSAC*No Desc. Maint.=true, Privacy=true, Method Category=End-to-end (E2E), MB used=1122023.06 | — | — | — | — | 0.3 | — | 0.4 | — | 0.3 | — | 0.4 | — | — | — | 0.15 | 0.21 | 0.05 | 0.13 | |
| GoMatchNo Desc. Maint.=true, Privacy=true, Method Category=Descriptor-free (DF), MB used=482023.06 | — | — | — | — | 0.64 | — | 8.14 | — | 4.77 | — | 9.94 | — | — | — | 0.25 | 2.83 | 0.48 | 3.35 | |
| HSCNetNo Desc. Maint.=true, Privacy=true, Method Category=End-to-end (E2E), MB used=5922023.06 | — | — | — | — | 0.3 | — | 0.3 | — | 0.3 | — | 0.3 | — | — | — | 0.18 | 0.19 | 0.06 | 0.09 | |
| HybridSCNo Desc. Maint.=false, Privacy=true, Method Category=Descriptor-based (DB), MB used=32023.06 | — | — | — | — | 0.59 | — | 1.01 | — | 0.54 | — | 0.49 | — | — | — | 0.81 | 0.75 | 0.19 | 0.5 | |
| MS-Trans.No Desc. Maint.=true, Privacy=true, Method Category=End-to-end (E2E), MB used=712023.06 | — | — | — | — | 1.47 | — | 2.39 | — | 3.07 | — | 3.99 | — | — | — | 0.83 | 1.81 | 0.86 | 1.62 | |
| SP+SGNo Desc. Maint.=false, Privacy=false, Method Category=Descriptor-based (DB), MB used=32152023.06 | — | — | — | — | 0.2 | — | 0.3 | — | 0.2 | — | 0.21 | — | — | — | 0.12 | 0.15 | 0.04 | 0.07 |