Research map / 3D vision / Depth and stereo
Princeton365: A Diverse Dataset with Accurate Camera Pose
Karhan Kayan, Stamatis Alexandropoulos, Rishabh Jain, Yiming Zuo, Erich Liang, Jia Deng
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 Bundle PnP refinement (ablation) | calibration-board ground truth · MoCap comparison (10 boards) | ATE (mm): 4.4 -> 2.1 | Improved |
| add pose graph optimization (ablation) | 16 boards, close filming · MoCap comparison | ATE (mm): 22 -> 7.5 | Improved |
| add snapping (coplanar boards assumption) (ablation) | 16 boards, close filming · MoCap comparison | ATE (mm): 22 -> 7.1 | Improved |
| add both pose graph optimization and snapping (ablation) | 16 boards, close filming · MoCap comparison | ATE (mm): 22 -> 3.0 | Improved |
| use multi-board PnP instead of nearest-board pose (ablation) | 8 boards, medium filming · MoCap comparison | ATE (mm): 8.1 -> 5.5 | Improved |
| build pose graph from close video, trajectory from medium video (C-M instead of M-M) (ablation) | 16 boards · MoCap comparison | ATE (mm): 3.2 -> 2.6 | Improved |
| use 16 boards instead of 10 without Bundle PnP (ablation) | calibration-board ground truth · MoCap comparison | ATE (mm): 4.4 -> 2.6 | Improved |
| use our board-based ground truth poses instead of COLMAP poses (ablation) | Nerfacto · Princeton365 NVS (Windows-Night) | PSNR: 20.30 -> 26.88 | Improved |
| use our board-based ground truth poses instead of COLMAP poses (ablation) | Splatfacto · Princeton365 NVS (Windows-Night) | PSNR: 23.28 -> 32.26 | Improved |
| use our ground truth instead of COLMAP for trajectory (ablation) | ground-truth collection · MoCap-validated sequences | mean ATE (mm): 11.4 -> 2.88 | Improved |
| use GridBoards instead of ChArUco boards (ablation) | ground-truth collection · MoCap board comparison | ATE (mm): 1.6 -> 1.1 | Improved |
Closest papers by meaning
- UniSim-SLAM: Feed-Forward SLAM with Unified Sim(3) Optimization (ECCV 2026 · similarity 0.88)
- MegaSaM: Accurate, Fast and Robust Structure and Motion from Casual Dynamic Videos (CVPR 2025 · similarity 0.86)
- VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifold (NeurIPS 2025 · similarity 0.86)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.86)
- SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos (CVPR 2025 · similarity 0.85)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · 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.
