Research map / 3D vision / Depth and stereo
Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometry
Nan Yang, Rui Wang, Jorg Stuckler, Daniel Cremers
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the virtual stereo term to windowed bundle adjustment (ablation) | DVSO (in) · KITTI odometry overall mean | t_rel: 10.76 -> 1.08 | Improved |
| add left-right disparity consistency check for point selection (ablation) | DVSO (in,vs) · KITTI odometry overall mean | t_rel: 1.08 -> 1.06 | No clear effect |
| add tuned virtual stereo baseline (ablation) | DVSO (in,vs,lr) · KITTI odometry overall mean | t_rel: 1.06 -> 0.77 | Improved |
| use StackNet depth instead of Godard et al. depth (ablation) | DVSO · KITTI odometry overall mean | t_rel: 1.51 -> 0.77 | Improved |
| add StackNet (cascaded ResidualNet) to SimpleNet (ablation) | depth network · KITTI Eigen 0-80m | RMSE: 4.785 -> 4.442 | Improved |
| add ResidualNet stacking | depth network · KITTI Eigen 1-50m | RMSE vs Kuznietsov: 3.518 -> 3.390 | Improved |
| train SimpleNet on Ko then Kr (self-supervised) (ablation) | SimpleNet · KITTI Eigen 0-80m | RMSE: 4.886 -> 4.817 | Improved |
| train SimpleNet on Kr then Ko only (ablation) | SimpleNet · KITTI Eigen 0-80m | RMSE: 4.886 -> 4.890 | No clear effect |
| add final fine-tuning on Ko after Ko->Kr (ablation) | SimpleNet · KITTI Eigen 0-80m | RMSE: 4.817 -> 4.785 | Improved |
| use semi-supervised sparse Stereo DSO depth plus self-supervision vs self-supervised Godard | depth network · KITTI Eigen 0-80m | RMSE: 4.935 -> 4.442 | Improved |
| use monocular DVSO instead of monocular DSO with Sim(3) alignment | odometry · KITTI odometry overall mean | t_rel: 65.64 -> 0.77 | Improved |
| use DVSO instead of Stereo DSO | odometry · KITTI odometry overall mean | t_rel: 0.84 -> 0.77 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Stereo DSO — Stereo DSO: Large-Scale Direct Sparse Visual Odometry With Stereo Cameras (ICCV 2017) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Unsupervised Learning of Monocular Depth Estimation and Visual Odometry With Deep Feature Reconstruction (CVPR 2018 · similarity 0.94)
- D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry (CVPR 2020 · similarity 0.92)
- Stereo DSO: Large-Scale Direct Sparse Visual Odometry With Stereo Cameras (ICCV 2017 · similarity 0.90)
- Learning Depth From Monocular Videos Using Direct Methods (CVPR 2018 · similarity 0.90)
- Generalizing to the Open World: Deep Visual Odometry With Online Adaptation (CVPR 2021 · similarity 0.89)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
