Camera Localization on 7 Scenes
0.02Average Position Error (m)DFNet+NeFeS
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DFNet+NeFeSMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.02 | 0.02 | 0.57 | 0.02 | 0.74 | 0.02 | 1.28 | 0.02 | 0.56 | 0.02 | 0.55 | 0.02 | 0.57 | 0.05 | 1.28 | 0.79 | — | — | |
| MarepoMethod Category=APR, Dataset-specific training time=15min / scene2024.12 | 0.03 | 0.02 | 1.24 | 0.02 | 1.39 | 0.02 | 2.03 | 0.03 | 1.26 | 0.04 | 1.48 | 0.04 | 1.71 | 0.06 | 1.67 | 1.54 | — | — | |
| CamNetMethod Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.04 | 0.04 | 1.73 | 0.03 | 1.74 | 0.05 | 1.98 | 0.04 | 1.62 | 0.04 | 1.64 | 0.04 | 1.63 | 0.04 | 1.51 | 1.69 | — | — | |
| Reloc3r-512Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.04 | 0.03 | 0.88 | 0.03 | 0.81 | 0.01 | 0.95 | 0.04 | 0.88 | 0.06 | 1.1 | 0.04 | 1.26 | 0.07 | 1.26 | 1.02 | — | — | |
| Reloc3r-224Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.05 | 0.03 | 0.99 | 0.04 | 1.13 | 0.02 | 1.23 | 0.05 | 0.88 | 0.07 | 1.14 | 0.05 | 1.23 | 0.12 | 2.25 | 1.26 | — | — | |
| PMNetMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.06 | 0.03 | 1.26 | 0.04 | 1.76 | 0.02 | 1.68 | 0.06 | 1.69 | 0.07 | 1.96 | 0.08 | 2.23 | 0.11 | 2.97 | 1.93 | — | — | |
| LENSMethod Category=APR, Dataset-specific training time=Days / scene2024.12 | 0.08 | 0.03 | 1.3 | 0.1 | 3.7 | 0.07 | 5.8 | 0.07 | 1.9 | 0.08 | 2.2 | 0.09 | 2.2 | 0.14 | 3.6 | 3 | — | — | |
| ExReNet (SUNCG)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.08 | 0.05 | 1.63 | 0.07 | 2.54 | 0.03 | 2.71 | 0.06 | 1.75 | 0.07 | 2.04 | 0.07 | 2.1 | 0.19 | 4.87 | 2.52 | — | — | |
| KV-TrackerResolution=350x2662025.12 | 0.08 | 0.091 | — | 0.042 | — | 0.054 | — | 0.065 | — | 0.142 | — | 0.038 | — | 0.128 | — | — | — | — | |
| AnchorNetMethod Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.09 | 0.06 | 3.89 | 0.15 | 10.3 | 0.08 | 10.9 | 0.09 | 5.15 | 0.1 | 2.97 | 0.08 | 4.68 | 0.1 | 9.26 | 6.74 | — | — | |
| ExReNet (SN)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.11 | 0.06 | 2.15 | 0.09 | 3.2 | 0.04 | 3.3 | 0.07 | 2.17 | 0.11 | 2.65 | 0.09 | 2.57 | 0.33 | 7.34 | 3.34 | — | — | |
| Map-free (Regress)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.13 | 0.09 | 2.66 | 0.13 | 4.54 | 0.11 | 4.81 | 0.11 | 2.77 | 0.16 | 3.11 | 0.14 | 3.48 | 0.18 | 4.7 | 3.72 | — | — | |
| Map-free (Match)Method Category=RPR (Unseen), Hybrid pose estimation=true, Dataset-specific training time=None2024.12 | 0.14 | 0.1 | 2.93 | 0.12 | 4.95 | 0.11 | 5.4 | 0.12 | 3.01 | 0.16 | 3.19 | 0.14 | 3.45 | 0.21 | 4.5 | 3.92 | — | — | |
| TTT3RResolution=512x3842025.12 | 0.143 | 0.154 | — | 0.124 | — | 0.097 | — | 0.196 | — | 0.228 | — | 0.136 | — | 0.063 | — | — | — | — | |
| Relpose-GNNMethod Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.16 | 0.08 | 2.7 | 0.21 | 7.5 | 0.13 | 8.7 | 0.15 | 4.1 | 0.15 | 3.5 | 0.19 | 3.7 | 0.22 | 6.5 | 5.2 | — | — | |
| MS-TransformerLearning paradigm=Multi-scene2021.03 | 0.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 7.28 | 1 | 1 | |
| AtLoc+Temporal Constraints=true2019.09 | 0.19 | 0.1 | 3.18 | 0.26 | 10.8 | 0.14 | 11.4 | 0.17 | 5.16 | 0.2 | 3.94 | 0.16 | 4.9 | 0.29 | 10.2 | 7.08 | — | — | |
| ImageNet+NCMMethod Category=RPR (Unseen), Hybrid pose estimation=true, Dataset-specific training time=None2024.12 | 0.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 4.3 | — | — | |
| AtLocInput type=Single-image, Temporal constraints=Without2019.09 | 0.2 | 0.1 | 4.07 | 0.25 | 11.4 | 0.16 | 11.8 | 0.17 | 5.34 | 0.21 | 4.37 | 0.23 | 5.42 | 0.26 | 10.5 | 7.56 | — | — | |
| AttLocLearning paradigm=Single-scene2021.03 | 0.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 7.56 | 2 | 2 | |
| MSPNLearning paradigm=Multi-scene2021.03 | 0.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 8.41 | 2 | 6 | |
| CUT3RResolution=512x3842025.12 | 0.205 | 0.297 | — | 0.218 | — | 0.115 | — | 0.356 | — | 0.249 | — | 0.118 | — | 0.079 | — | — | — | — | |
| MapNetTemporal Constraints=true2019.09 | 0.21 | 0.08 | 3.25 | 0.27 | 11.7 | 0.18 | 13.3 | 0.17 | 5.15 | 0.22 | 4.02 | 0.23 | 4.93 | 0.3 | 12.1 | 7.77 | — | — | |
| MapNetLearning paradigm=Single-scene2021.03 | 0.21 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 7.78 | 4 | 3 | |
| Relative PN (7S)Method Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.21 | 0.13 | 6.46 | 0.26 | 12.72 | 0.14 | 12.34 | 0.21 | 7.35 | 0.24 | 6.35 | 0.24 | 8.03 | 0.27 | 11.82 | 9.3 | — | — | |
| NC-EssNet (7S)Method Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.21 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 7.5 | — | — | |
| RelocNet (7S)Method Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.21 | 0.12 | 4.14 | 0.26 | 10.4 | 0.14 | 10.5 | 0.18 | 5.32 | 0.26 | 4.17 | 0.23 | 5.08 | 0.28 | 7.53 | 6.73 | — | — | |
| EssNet (7S)Method Category=RPR (Seen), Dataset-specific training time=Hours2024.12 | 0.22 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 8.03 | — | — | |
| HourglassInput type=Single-image, Temporal constraints=Without2019.09 | 0.23 | 0.15 | 6.17 | 0.27 | 10.8 | 0.19 | 11.6 | 0.21 | 8.48 | 0.25 | 7.01 | 0.27 | 10.2 | 0.29 | 12.5 | 9.53 | — | — | |
| PoseNet17Input type=Single-image, Temporal constraints=Without2019.09 | 0.23 | 0.13 | 4.48 | 0.27 | 11.3 | 0.17 | 13 | 0.19 | 5.55 | 0.26 | 4.75 | 0.23 | 5.35 | 0.35 | 12.4 | 8.12 | — | — | |
| GeoPoseNetLearning paradigm=Single-scene2021.03 | 0.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 8.12 | 5 | 5 | |
| IRPNetLearning paradigm=Single-scene2021.03 | 0.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 8.49 | 5 | 7 | |
| PoseNet-LearnableLearning paradigm=Single-scene2021.03 | 0.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 7.87 | 7 | 4 | |
| VidLocTemporal Constraints=true2019.09 | 0.25 | 0.18 | — | 0.26 | — | 0.14 | — | 0.26 | — | 0.36 | — | 0.31 | — | 0.26 | — | — | — | — | |
| RelocNet (SN)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.29 | 0.21 | 10.9 | 0.32 | 11.8 | 0.15 | 13.4 | 0.31 | 10.3 | 0.4 | 10.9 | 0.33 | 10.3 | 0.33 | 11.4 | 11.3 | — | — | |
| PoseNet Spatial LSTMInput type=Single-image, Temporal constraints=Without2019.09 | 0.31 | 0.24 | 5.77 | 0.34 | 11.9 | 0.21 | 13.7 | 0.3 | 8.08 | 0.33 | 7 | 0.37 | 8.83 | 0.4 | 13.7 | 9.85 | — | — | |
| LSTM-PNLearning paradigm=Single-scene2021.03 | 0.31 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 9.86 | 8 | 9 | |
| GPoseNetLearning paradigm=Single-scene2021.03 | 0.31 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 9.95 | 8 | 8 | |
| Relative PN (U)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.36 | 0.31 | 15.05 | 0.4 | 19 | 0.24 | 22.15 | 0.38 | 14.14 | 0.44 | 18.24 | 0.41 | 16.51 | 0.35 | 23.55 | 18.38 | — | — | |
| Point3RResolution=224x2242025.12 | 0.439 | 0.427 | — | 0.28 | — | 0.389 | — | 0.436 | — | 0.644 | — | 0.502 | — | 0.398 | — | — | — | — | |
| PoseNetLearning paradigm=Single-scene2021.03 | 0.44 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 10.4 | 10 | 11 | |
| PoseNetInput type=Single-image, Temporal constraints=Without2019.09 | 0.45 | 0.32 | 6.6 | 0.47 | 14 | 0.3 | 12.2 | 0.48 | 7.24 | 0.49 | 8.12 | 0.58 | 8.34 | 0.48 | 13.1 | 9.94 | — | — | |
| Bayesian PoseNetInput type=Single-image, Temporal constraints=Without2019.09 | 0.47 | 0.37 | 7.24 | 0.43 | 13.7 | 0.31 | 12 | 0.48 | 8.04 | 0.61 | 7.08 | 0.58 | 7.54 | 0.48 | 13.1 | 9.81 | — | — | |
| BayesianPNLearning paradigm=Single-scene2021.03 | 0.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 9.81 | 11 | 8 | |
| NC-EssNet (CL)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.48 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 32.97 | — | — | |
| EssNet (CL)Method Category=RPR (Unseen), Dataset-specific training time=None2024.12 | 0.57 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.06 | — | — | |
| Active SearchMethod Type=Sparse, Scene-specific training=false, Input Context=Single frame2021.03 | — | 0.04 | 1.96 | 0.03 | 1.53 | 0.02 | 1.45 | 0.09 | 3.61 | 0.08 | 3.1 | 0.07 | 3.37 | 0.03 | 2.22 | — | — | — | |
| DSACMethod Type=Dense, Scene-specific training=true, Input Context=Single frame2021.03 | — | 0.02 | 0.7 | 0.03 | 1 | 0.02 | 1.3 | 0.03 | 1 | 0.05 | 1.3 | 0.05 | 1.5 | 1.9 | 49.4 | — | — | — | |
| DSAC++Method Type=Dense, Scene-specific training=true, Input Context=Single frame2021.03 | — | 0.02 | 0.5 | 0.02 | 0.9 | 0.01 | 0.8 | 0.03 | 0.7 | 0.04 | 1.1 | 0.04 | 1.1 | 0.09 | 2.6 | — | — | — | |
| DSMMethod Type=Dense, Scene-specific training=false, Input Context=Single frame2021.03 | — | 0.02 | 0.71 | 0.02 | 0.85 | 0.01 | 0.85 | 0.03 | 0.84 | 0.04 | 1.16 | 0.04 | 1.17 | 0.05 | 1.32 | — | — | — | |
| DSMMethod Type=Dense, Scene-specific training=false, Input Context=Video2021.03 | — | 0.02 | 0.68 | 0.02 | 0.8 | 0.01 | 0.8 | 0.03 | 0.78 | 0.04 | 1.11 | 0.03 | 1.11 | 0.04 | 1.16 | — | — | — | |
| HLocMethod Type=Sparse, Scene-specific training=false, Input Context=Single frame2021.03 | — | 0.02 | 0.79 | 0.02 | 0.87 | 0.02 | 0.92 | 0.03 | 0.91 | 0.05 | 1.12 | 0.04 | 1.25 | 0.06 | 1.62 | — | — | — | |
| InLocMethod Type=Sparse, Scene-specific training=false, Input Context=Single frame2021.03 | — | 0.03 | 1.05 | 0.03 | 1.07 | 0.02 | 0.16 | 0.03 | 1.05 | 0.05 | 1.55 | 0.04 | 1.31 | 0.09 | 2.47 | — | — | — | |
| KFNetMethod Type=Dense, Scene-specific training=true, Input Context=Single frame2021.03 | — | 0.02 | 0.65 | 0.02 | 0.9 | 0.01 | 0.82 | 0.03 | 0.69 | 0.04 | 1.02 | 0.04 | 1.16 | 0.03 | 0.94 | — | — | — | |
| SANetMethod Type=Dense, Scene-specific training=false, Input Context=Single frame2021.03 | — | 0.03 | 0.88 | 0.03 | 1.1 | 0.02 | 1.48 | 0.03 | 1.03 | 0.05 | 1.33 | 0.04 | 1.4 | 0.16 | 4.59 | — | — | — |