Camera Pose Regression on Oxford RobotCar (Average)
8.03Mean Translation Error (m)CoordiNet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CoordiNetLoss=Homosc., input=1, Backbone=ResNet342021.03 | 8.03 | 3.8 | 3.32 | 0.97 | |
| CoordiNetLoss=Heterosc., input=1, Backbone=ResNet342021.03 | 9.56 | 3.59 | 2.91 | 1 | |
| AD-MapNetLoss=Homosc., input=2, Backbone=ResNet342021.03 | 12.82 | 3.79 | — | — | |
| AtLoc+Loss=Homosc., input=2, Backbone=ResNet342021.03 | 14.27 | 4.88 | 5.23 | 1.74 | |
| AtLocLoss=Homosc., input=1, Backbone=ResNet342021.03 | 19.17 | 8.52 | 8.23 | 3.69 | |
| AD-PoseNetLoss=Homosc., input=1, Backbone=ResNet342021.03 | 20.11 | 4.93 | — | — | |
| MapNetLoss=Homosc., input=2, Backbone=ResNet342021.03 | 26.95 | 6.86 | — | — | |
| PoseNetLoss=Homosc., input=1, Backbone=ResNet342021.03 | 27.26 | 6.99 | — | — |