Research map / 3D vision / Human pose estimation
Towards Accurate Active Camera Localization
Qihang Fang, Yingda Yin, Qingnan Fan, Fei Xia, Siyan Dong, Sheng Wang, Jue Wang, Leonidas J. Guibas, Baoquan Chen
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 exploration reward (ablation) | Ours · ACL-synthetic (5cm, 5deg) | accuracy (%): 72.50 -> 83.05 | Improved |
| add exploration reward (ablation) | Ours · ACL-real (5cm, 5deg) | accuracy (%): 73.00 -> 82.40 | Improved |
| add camera-driven scene map (with exploration reward removed) (ablation) | Ours · ACL-synthetic (5cm, 5deg) | accuracy (%): 67.65 -> 72.50 | Improved |
| add world-driven scene map (with exploration reward removed) (ablation) | Ours · ACL-synthetic (5cm, 5deg) | accuracy (%): 66.40 -> 72.50 | Improved |
| add camera-driven scene map (with exploration reward removed) (ablation) | Ours · ACL-real (5cm, 5deg) | accuracy (%): 70.60 -> 73.00 | Improved |
| add world-driven scene map (with exploration reward removed) (ablation) | Ours · ACL-real (5cm, 5deg) | accuracy (%): 67.40 -> 73.00 | Improved |
| active learned policy instead of turn-around heuristic | Ours · ACL-synthetic (5cm, 5deg) | accuracy (%): 25.00 -> 83.05 | Improved |
| enable active movement instead of no-movement | DSAC passive localizer · ACL-synthetic (5cm, 5deg) | accuracy (%): 14.90 -> 83.05 | Improved |
| learned RL policy instead of ANL at fine scale | Ours · ACL-synthetic (5cm, 5deg) | accuracy (%): 3.25 -> 83.05 | Improved |
| ours instead of ANL at coarse scale | Ours · ACL-synthetic (20cm, 90deg) | accuracy (%): 87.30 -> 88.75 | No clear effect |
| halve the RGB-D stream (sparse data) | Ours · ACL-real (5cm, 5deg) | accuracy (%): 82.40 -> 76.40 | Hurt |
| halve the RGB-D stream (sparse data) | Scene-descent · ACL-synthetic (5cm, 5deg) | accuracy (%): 57.65 -> 19.90 | Hurt |
Closest papers by meaning
- Active Camera Relocalization from a Single Reference Image without Hand-Eye Calibration (TPAMI 2018 · similarity 0.88)
- Pose Correction for Highly Accurate Visual Localization in Large-Scale Indoor Spaces (ICCV 2021 · similarity 0.88)
- Combining Absolute and Semi-Generalized Relative Poses for Visual Localization (IJCV 2026 · similarity 0.87)
- Reloc3r: Large-Scale Training of Relative Camera Pose Regression for Generalizable, Fast, and Accurate Visual Localization (CVPR 2025 · similarity 0.85)
- Cycle-Sync: Robust Global Camera Pose Estimation through Enhanced Cycle-Consistent Synchronization (NeurIPS 2025 · similarity 0.85)
- Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping (NeurIPS 2025 · similarity 0.84)
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.
