AI Paper Atlas

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

ECCV 2018 · Learning theory and algorithms › Combinatorial and search optimization · approach: Globally optimal geometric optimization · 83 citations (OpenAlex)

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 changedWhereMeasuredEffect
increase learned optimization iterations from initialization to 15 steps (ablation)LS-Net · RGB-DL1-inv: 0.260 -> 0.019Improved
increase learned optimization iterations from initialization to 15 steps (ablation)LS-Net · RGB-Dsc-inv: 0.360 -> 0.09Improved
increase learned optimization iterations from initialization to 15 steps (ablation)LS-Net · RGB-Drotation error: 2.290 -> 1.01Improved
increase learned optimization iterations from initialization to 15 steps (ablation)LS-Net · RGB-Dtranslation error: 27.40 -> 22.14Improved
increase optimization steps from 5 to 15 (ablation)LS-Net · RGB-Dsc-inv: 0.15 -> 0.09Improved
increase optimization steps from 5 to 10 (ablation)LS-Net · RGB-Drotation error: 2.11 -> 1.23Improved
increase optimization steps from 5 to 15 (ablation)LS-Net · RGB-DL1-rel: 0.308 -> 0.301No clear effect
use learned LSTM optimizer instead of Levenberg-MarquardtLS-Net · synthetic curve fittingconvergence speedImproved
use learned network update instead of sparse matrix inversion in LMLS-Net · depth and motion (128x96 image)time per step: 532ms -> 25msImproved
use LS-Net instead of DeMoN iterative networkLS-Net · model sizeparameters: 45,753,883 -> 11,438,470Improved
use LS-Net instead of DeMoNLS-Net · inferencespeed: 1.5x fasterImproved
use LS-Net instead of DeMoNLS-Net · MVSL1-inv: 0.047 -> 0.051Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy