Research map / Architectures and efficiency / Quantization
Square Root Bundle Adjustment for Large-Scale Reconstruction
Check 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 |
|---|---|---|---|
| Replace Schur complement reduction with QR-based nullspace marginalization | Bundle adjustment on BAL · BAL datasets | runtime: 42% faster | Helped |
| Use single precision instead of double precision (ablation) | √BA · BAL datasets | runtime: around twice as fast | Helped |
| Use single precision instead of double precision (ablation) | √BA · BAL datasets | accuracy | No clear effect |
| Use single-precision explicit Schur complement instead of double precision (ablation) | Custom explicit Schur complement solver · BAL datasets | accuracy | Hurt |
| Use single-precision explicit Schur complement instead of double precision (ablation) | Custom explicit Schur complement solver · BAL datasets | runtime | Mixed |
| Use single-precision explicit Schur complement (ablation) | Custom explicit Schur complement solver · BAL datasets | problems requiring backtracking due to numerical indefiniteness: 84 out of the 97 problems | Hurt |
| Use double-precision explicit Schur complement (ablation) | Custom explicit Schur complement solver · BAL datasets | problems with numerical indefiniteness: 7 of the problems | Hurt |
| Use square root bundle adjustment instead of explicit Schur complement | Bundle adjustment on BAL · BAL datasets | problems with numerical indefiniteness | Helped |
| Use landmark-block QR storage and marginalization | √BA · final13682 BAL | memory feasibility | Hurt |
| Use implicit Schur complement rather than √BA | Bundle adjustment on BAL · Very large problems | runtime scalability | Mixed |
| Use √BA rather than competing solvers | Bundle adjustment on BAL · Medium to large problems | runtime | Helped |
Closest papers by meaning
- Power Bundle Adjustment for Large-Scale 3D Reconstruction (CVPR 2023 · similarity 0.89)
- A Consensus-Based Framework for Distributed Bundle Adjustment (CVPR 2016 · similarity 0.85)
- Matrix-Free Shared Intrinsics Bundle Adjustment (CVPR 2025 · similarity 0.85)
- Efficient Second-Order Plane Adjustment (CVPR 2023 · similarity 0.83)
- BA-T: An Iterative Transformer for Two-View Bundle Adjustment (NeurIPS 2026 · similarity 0.82)
- Schwarz–Schur Involution: Lightspeed Differentiable Sparse Linear Solvers (ICML 2025 · similarity 0.80)
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.