Urban Localization on RobotCar Seasons v2
9.8Recall (0.25m/2°)TransVPR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TransVPRRe-ranking=true, Training Dataset=MSLS2023.03 | 9.8 | 34.7 | 80 | |
| Patch NetVLADPCA=Yes, Dimension=4096, Re-ranking=true, Training Dataset=MSLS2023.03 | 9.6 | 35.3 | 90.9 | |
| SP-SuperGlueRe-ranking=true, Training Dataset=MSLS2023.03 | 9.5 | 35.4 | 85.4 | |
| ResNet152-GeM-GCLPCA=Yes, Dimension=2048, Re-ranking=false, Training Dataset=MSLS2023.03 | 6 | 21.6 | 72.5 | |
| VGG-GeM-GCLPCA=Yes, Dimension=512, Re-ranking=false, Training Dataset=MSLS2023.03 | 5.4 | 21.9 | 69.2 | |
| NetVLAD 16PCA=Yes, Dimension=4096, Re-ranking=false, Training Dataset=MSLS2023.03 | 4.8 | 17.9 | 65.3 | |
| ResNet50-GeM-GCLPCA=Yes, Dimension=1024, Re-ranking=false, Training Dataset=MSLS2023.03 | 4.7 | 20.2 | 70 | |
| ResNeXt-GeM-GCLPCA=Yes, Dimension=1024, Re-ranking=false, Training Dataset=MSLS2023.03 | 4.7 | 21 | 74.7 | |
| NetVLAD 64PCA=Yes, Dimension=4096, Re-ranking=false, Training Dataset=MSLS2023.03 | 4.2 | 18 | 68.1 | |
| VGG-GeM-GCLPCA=No, Dimension=512, Re-ranking=false, Training Dataset=MSLS2023.03 | 3.7 | 15.8 | 59.7 | |
| NetVLAD-GCLPCA=Yes, Dimension=4096, Re-ranking=false, Training Dataset=MSLS2023.03 | 3.4 | 14.2 | 58.8 | |
| NetVLAD-GCLPCA=No, Dimension=32768, Re-ranking=false, Training Dataset=MSLS2023.03 | 3.3 | 14.1 | 58.2 | |
| TransVPRRe-ranking=false, Training Dataset=MSLS2023.03 | 2.9 | 11.4 | 58.6 | |
| ResNet50-GeM-GCLPCA=No, Dimension=2048, Re-ranking=false, Training Dataset=MSLS2023.03 | 2.9 | 14 | 58.8 | |
| ResNet152-GeM-GCLPCA=No, Dimension=2048, Re-ranking=false, Training Dataset=MSLS2023.03 | 2.9 | 13.1 | 63.5 | |
| ResNeXt-GeM-GCLPCA=No, Dimension=2048, Re-ranking=false, Training Dataset=MSLS2023.03 | 2.7 | 13.4 | 65.2 | |
| DELGRe-ranking=true, Training Dataset=MSLS2023.03 | 2.2 | 8.4 | 76.8 | |
| NetVLAD 64PCA=No, Dimension=32768, Re-ranking=false, Training Dataset=MSLS2023.03 | 2 | 9.2 | 45.5 | |
| NetVLAD 16PCA=No, Dimension=8192, Re-ranking=false, Training Dataset=MSLS2023.03 | 1.8 | 9.2 | 48.4 |