Research map / 3D vision / Depth and stereo
A Global Occlusion-Aware Approach to Self-Supervised Monocular Visual Odometry
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 proposed intra-stage augmented attention (ablation) | Baseline depth model · KITTI depth ablation | Abs Rel: 0.163 -> 0.134 | Helped |
| Add cascaded across-stage attention (ablation) | Depth model with proposed ISAA · KITTI depth ablation | Abs Rel: 0.134 -> 0.119 | Helped |
| Add the adversarial global constraint (ablation) | Depth model with ISAA and CASA · KITTI depth ablation | Abs Rel: 0.119 -> 0.109 | Helped |
| Replace proposed ISAA with inner latent-feature self-attention (ablation) | Baseline depth model · KITTI depth ablation | Abs Rel: 0.163 -> 0.153 | Helped |
| Replace proposed ISAA with stage-wise decoder-feature self-attention (ablation) | Baseline depth model · KITTI depth ablation | Abs Rel: 0.163 -> 0.147 | Helped |
| Replace proposed ISAA with traditional non-local attention on decoder features (ablation) | Baseline depth model · KITTI depth ablation | Abs Rel: 0.163 -> 0.140 | Helped |
| Use proposed ISAA instead of the inner, stage-wise, or traditional non-local attention variants (ablation) | Baseline depth model · KITTI depth ablation | Abs Rel: 0.153/0.147/0.140 -> 0.134 | Helped |
| Use the proposed global occlusion-aware method | Self-supervised depth estimation · KITTI raw (Eigen split) | Abs Rel: 0.111 -> 0.109 | Helped |
| Use the proposed global occlusion-aware method | Self-supervised depth estimation · KITTI raw (Eigen split) | δ < 1.25: 0.885 -> 0.881 | Hurt |
| Estimate pose with the proposed method using two frames rather than GANVO's five-frame setup | Monocular visual odometry · KITTI odometry Sequence 09 | ATE: 0.009 ± 0.005 -> 0.013 ± 0.006 | Hurt |
| Estimate pose with the proposed method using two frames rather than GANVO's five-frame setup | Monocular visual odometry · KITTI odometry Sequence 10 | ATE: 0.010 ± 0.013 -> 0.012 ± 0.006 | Hurt |
Closest papers by meaning
- LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry (CVPR 2024 · similarity 0.88)
- ZeroVO: Visual Odometry with Minimal Assumptions (CVPR 2025 · similarity 0.87)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.87)
- CorrBEV: Multi-View 3D Object Detection by Correlation Learning with Multi-modal Prototypes (CVPR 2025 · similarity 0.86)
- Leveraging Consistent Spatio-Temporal Correspondence for Robust Visual Odometry (AAAI 2025 · similarity 0.86)
- SelfOcc: Self-Supervised Vision-Based 3D Occupancy Prediction (CVPR 2024 · 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.