BA-Net: Dense Bundle Adjustment Network
About
This paper introduces a network architecture to solve the structure-from-motion (SfM) problem via feature-metric bundle adjustment (BA), which explicitly enforces multi-view geometry constraints in the form of feature-metric error. The whole pipeline is differentiable so that the network can learn suitable features that make the BA problem more tractable. Furthermore, this work introduces a novel depth parameterization to recover dense per-pixel depth. The network first generates several basis depth maps according to the input image and optimizes the final depth as a linear combination of these basis depth maps via feature-metric BA. The basis depth maps generator is also learned via end-to-end training. The whole system nicely combines domain knowledge (i.e. hard-coded multi-view geometry constraints) and deep learning (i.e. feature learning and basis depth maps learning) to address the challenging dense SfM problem. Experiments on large scale real data prove the success of the proposed method.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Depth Estimation | KITTI (Eigen split) | RMSE3.64 | 276 | |
| Depth Estimation | ScanNet (test) | Abs Rel0.161 | 65 | |
| Depth Estimation | Sun3D (test) | Abs Rel6 | 22 | |
| Depth Estimation | Scenes11 (test) | L1 Relative Error0.13 | 12 | |
| Pose Estimation | MVS DeMoN version (test) | Rot Error3.499 | 8 | |
| Pose Estimation | Sun3D (test) | Rotation Error1.729 | 8 | |
| Pose Estimation | Scenes11 (test) | Rotation Error3.499 | 8 | |
| Video Depth Estimation | ScanNet (in-domain) | Abs Rel0.091 | 8 | |
| Two-view Depth Estimation | ScanNet (test) | Abs Rel0.161 | 8 | |
| Depth Estimation | MVS DeMoN (test) | L1-rel0.08 | 7 |