Research map / Learning theory and algorithms / Combinatorial and search optimization
Learning to Solve Nonlinear Least Squares for Monocular Stereo
Ronald Clark, Michael Bloesch, Jan Czarnowski, Stefan Leutenegger, Andrew J. Davison
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 |
|---|---|---|---|
| increase learned optimization iterations from initialization to 15 steps (ablation) | LS-Net · RGB-D | L1-inv: 0.260 -> 0.019 | Improved |
| increase learned optimization iterations from initialization to 15 steps (ablation) | LS-Net · RGB-D | sc-inv: 0.360 -> 0.09 | Improved |
| increase learned optimization iterations from initialization to 15 steps (ablation) | LS-Net · RGB-D | rotation error: 2.290 -> 1.01 | Improved |
| increase learned optimization iterations from initialization to 15 steps (ablation) | LS-Net · RGB-D | translation error: 27.40 -> 22.14 | Improved |
| increase optimization steps from 5 to 15 (ablation) | LS-Net · RGB-D | sc-inv: 0.15 -> 0.09 | Improved |
| increase optimization steps from 5 to 10 (ablation) | LS-Net · RGB-D | rotation error: 2.11 -> 1.23 | Improved |
| increase optimization steps from 5 to 15 (ablation) | LS-Net · RGB-D | L1-rel: 0.308 -> 0.301 | No clear effect |
| use learned LSTM optimizer instead of Levenberg-Marquardt | LS-Net · synthetic curve fitting | convergence speed | Improved |
| use learned network update instead of sparse matrix inversion in LM | LS-Net · depth and motion (128x96 image) | time per step: 532ms -> 25ms | Improved |
| use LS-Net instead of DeMoN iterative network | LS-Net · model size | parameters: 45,753,883 -> 11,438,470 | Improved |
| use LS-Net instead of DeMoN | LS-Net · inference | speed: 1.5x faster | Improved |
| use LS-Net instead of DeMoN | LS-Net · MVS | L1-inv: 0.047 -> 0.051 | Hurt |
Closest papers by meaning
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (CVPR 2020 · similarity 0.85)
- Single View Stereo Matching (CVPR 2018 · similarity 0.85)
- A Lightweight Neural Network for Monocular View Generation With Occlusion Handling (TPAMI 2019 · similarity 0.85)
- Bilateral Cyclic Constraint and Adaptive Regularization for Unsupervised Monocular Depth Prediction (CVPR 2019 · similarity 0.85)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.85)
- Certifiably Optimal Outlier-Robust Geometric Perception: Semidefinite Relaxations and Scalable Global Optimization (TPAMI 2022 · 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.
