Research map / Graph learning / Graph self-supervised learning
Learning Multi-View Camera Relocalization With Graph Neural Networks
Fei Xue, Xin Wu, Shaojun Cai, Junqiu Wang
Check 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 |
|---|---|---|---|
| use graph-based multi-view relocalization instead of LsG | camera relocalization · 7Scenes | orientation error: 7.47° -> 6.33° | Improved |
| use graph-based multi-view relocalization instead of MLFBPPose | camera relocalization · Cambridge | position error: 1.20m -> 1.12m | Improved |
| use graph-based multi-view relocalization instead of PoseNet17 | camera relocalization · Cambridge | orientation error: 2.85° -> 2.40° | Improved |
| use graph-based multi-view relocalization instead of LsG | camera relocalization · Oxford RobotCar | position error: 25.84m -> 17.77m | Improved |
| use graph-based multi-view relocalization instead of LsG | camera relocalization · Oxford RobotCar | orientation error: 4.99° -> 4.03° | Improved |
| replace standard GNNs with redesigned GNNs that retain 3D feature tensors (ablation) | Ours on Oxford RobotCar · Oxford RobotCar | position error: 35.18m -> 20.30m | Improved |
| replace standard GNNs with redesigned GNNs that retain 3D feature tensors (ablation) | Ours on Oxford RobotCar · Oxford RobotCar | orientation error: 8.53° -> 5.37° | Improved |
| add a pose graph constraint (ablation) | Ours with the level 4 GNN block on Oxford RobotCar · Oxford RobotCar | position error: 20.30m -> 19.44m | Improved |
| add a pose graph constraint (ablation) | Ours with the level 4 GNN block on Oxford RobotCar · Oxford RobotCar | orientation error: 5.37° -> 4.75° | Improved |
| add level 3 GNN processing to the level 4 model with pose graph constraint (ablation) | Ours on Oxford RobotCar · Oxford RobotCar | position error: 19.44m -> 18.66m | Improved |
| add level 3 GNN processing to the level 4 model with pose graph constraint (ablation) | Ours on Oxford RobotCar · Oxford RobotCar | orientation error: 4.75° -> 5.02° | Hurt |
| add level 2 GNN processing to the level 4 and 3 model with pose graph constraint (ablation) | Ours on Oxford RobotCar · Oxford RobotCar | position error: 18.66m -> 17.58m | Improved |
Closest papers by meaning
- PoGO-Net: Pose Graph Optimization With Graph Neural Networks (ICCV 2021 · similarity 0.88)
- Accelerated Coordinate Encoding: Learning to Relocalize in Minutes Using RGB and Poses (CVPR 2023 · similarity 0.88)
- Learning Global Camera Poses from Noisy View-Graphs for Structure from Motion (ECCV 2026 · similarity 0.88)
- Reloc3r: Large-Scale Training of Relative Camera Pose Regression for Generalizable, Fast, and Accurate Visual Localization (CVPR 2025 · similarity 0.87)
- Scene Coordinate Reconstruction: Posing of Image Collections via Incremental Learning of a Relocalizer (ECCV 2024 · similarity 0.87)
- Learning to Filter Outlier Edges in Global SfM (CVPR 2025 · similarity 0.85)
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.
