Research map / Detection and segmentation / Tracking
Deep Visual Odometry With Adaptive Memory
Fei Xue, Xin Wang, Junqiu Wang, Hongbin Zha
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 Remembering and Refining modules to Tracking (ablation) | Ours (tracking) · TUM-RGBD fr2/pioneer_360 | translational RMSE: 0.241 -> 0.056 | Improved |
| add Remembering and Refining modules to Tracking (ablation) | Ours (tracking) · TUM-RGBD fr3/nstr_ntex_near_loop | translational RMSE: 0.371 -> 0.123 | Improved |
| add temporal attention (ablation) | Ours (w/o temp atten) · TUM-RGBD fr2/pioneer_360 | translational RMSE: 0.1338 -> 0.056 | Improved |
| add spatial attention (ablation) | Ours (w/o spat atten) · TUM-RGBD fr2/pioneer_360 | translational RMSE: 0.076 -> 0.056 | Improved |
| add spatial attention (ablation) | Ours (w/o spat atten) · TUM-RGBD fr3/nstr_ntex_near_loop | translational RMSE: 0.157 -> 0.123 | Improved |
| add temporal attention (ablation) | Ours (w/o temp atten) · TUM-RGBD fr2/desk | translational RMSE: 0.164 -> 0.153 | Improved |
| increase sequence length from 5 to 11 frames (ablation) | Ours · KITTI | t_rel (%): 4.84 -> 3.47 | Improved |
| increase sequence length from 5 to 11 frames (ablation) | Ours · KITTI | r_rel: 2.65 -> 1.75 | Improved |
| increase sequence length from 5 to 11 frames (ablation) | Ours · TUM-RGBD fr2/desk | translational RMSE: 0.230 -> 0.153 | Improved |
| increase sequence length from 7 to 11 frames (ablation) | Ours · TUM-RGBD fr2/pioneer_360 | translational RMSE: 0.054 -> 0.056 | No clear effect |
| increase memory to 11 frames (ablation) | Ours · Running time | ms per frame: 13.1 -> 17.8 | Hurt |
| add Remembering and Refining to Tracking (ablation) | Ours (tracking) · Running time | ms per frame: 12.4 -> 17.8 | Hurt |
Closest papers by meaning
- Beyond Tracking: Selecting Memory and Refining Poses for Deep Visual Odometry (CVPR 2019 · similarity 0.96)
- Generalizing to the Open World: Deep Visual Odometry With Online Adaptation (CVPR 2021 · similarity 0.90)
- LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry (CVPR 2024 · similarity 0.89)
- Adaptive VIO: Deep Visual-Inertial Odometry with Online Continual Learning (CVPR 2024 · similarity 0.88)
- Leveraging Consistent Spatio-Temporal Correspondence for Robust Visual Odometry (AAAI 2025 · similarity 0.88)
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · similarity 0.86)
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.
