Research map / 3D vision / Depth and stereo
Stochastic Bundle Adjustment for Efficient and Scalable 3D Reconstruction
Lei Zhou, Zixin Luo, Mingmin Zhen, Tianwei Shen, Shiwei Li, Zhuofei Huang, Tian Fang, Long Quan
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 |
|---|---|---|---|
| use STBA instead of DBACC for distributed BA | distributed bundle adjustment · Large-Scale LS-1 | mean iteration time (s): 912.5 -> 71.0 | Improved |
| use STBA instead of DBACC | distributed BA · Large-Scale LS-4 | mean iteration time (s): 988.1 -> 71.2 | Improved |
| use STBA instead of DBACC | distributed BA · Large-Scale LS-1 | Jacobian evaluations: 1011 -> 49 | Improved |
| use STBA instead of DBACC | distributed BA · Large-Scale LS-2 | median reprojection error (pixel): 0.766 -> 0.783 | Hurt |
| use STBA instead of DBACC | distributed BA · Large-Scale LS-3 | median reprojection error (pixel): 1.083 -> 1.056 | Improved |
| add steepest descent correction (ablation) | STBA · 1DSfM and KITTI convergence | convergence speed | Improved |
| use stochastic clustering instead of fixed clustering (ablation) | STBA · 1DSfM and KITTI, tau=0.001 | performance profile | Improved |
| use randomized stochastic graph clustering instead of greedy deterministic Louvain (NSGC) (ablation) | STBA · 1DSfM and KITTI | normalized final loss | Improved |
| increase maximum cluster size from 1 to 25 (ablation) | STBA · 1DSfM and KITTI | final loss | Improved |
| use stochastic graph clustering instead of KMeans or NCut (ablation) | STBA · 1DSfM and KITTI | normalized final loss | Improved |
| increase maximum cluster size from 25 to 200 (ablation) | STBA with NSGC · 1DSfM and KITTI | final loss | Hurt |
| use STBA instead of LM/Dogleg at tau 0.1 and 0.01 | STBA · 1DSfM and KITTI | time to threshold | Improved |
Closest papers by meaning
- A Consensus-Based Framework for Distributed Bundle Adjustment (CVPR 2016 · similarity 0.89)
- Distributed Very Large Scale Bundle Adjustment by Global Camera Consensus (ICCV 2017 · similarity 0.89)
- Stable and Scalable Bundle Adjustment of Holistic 3D Structures (ECCV 2026 · similarity 0.89)
- Parallel Rigidity Matters for Bundle Adjustment (CVPR 2026 · similarity 0.89)
- Matrix-Free Shared Intrinsics Bundle Adjustment (CVPR 2025 · similarity 0.84)
- BA-T: An Iterative Transformer for Two-View Bundle Adjustment (NeurIPS 2026 · 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.
