AI Paper Atlas

Research map / 3D vision / Depth and stereo

LDL: Line Distance Functions for Panoramic Localization

Junho Kim, Changwoon Choi, Hojun Jang, Young Min Kim

ICCV 2023 · 3D vision › Depth and stereo · approach: Scene coordinate regression for localization · 3 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
add distance function decomposition along principal directions (ablation)LDL · OmniScenes Extremeaccuracy: 0.37 -> 0.92Improved
use 42 query points instead of 10 (ablation)LDL · OmniScenes Extremeaccuracy: 0.77 -> 0.92Improved
increase query points from 42 to 84 (ablation)LDL · OmniScenes Extremeaccuracy: 0.92 -> 0.95Mixed
use inlier-counting robust loss instead of L1/L2/Huber/Median (ablation)LDL · OmniScenes Wedding Hallaccuracy: 0.55 -> 0.68Improved
sample query points uniformly on sphere instead of along lines (ablation)LDL · Stanford 2D-3D-S (Total)accuracy: 0.66 -> 0.83Improved
use noisier SfM line maps instead of 3D scan lines (ablation)LDL · OmniScenes Rooms 3,5accuracy: 0.98 -> 0.85Hurt
line-based feature filtering for privacy (ablation)LDL refinement · Stanford 2D-3D-S (Total)accuracy: 0.83 -> 0.77Hurt
use line distance functions under synthetic illumination changesLDL vs CPO · OmniScenes Room 3 Extreme (gamma)accuracy: 0.00 -> 0.82Improved
line distance functions in place of Chamfer distanceline-based pose selection · Stanford 2D-3D-S (Total)accuracy: 0.39 -> 0.83Improved
LDL vs structure-based (NetVLAD) on clean Stanford dataLDL · Stanford 2D-3D-S (Total)accuracy: 0.85 -> 0.83No clear effect
run pose search on GPU instead of CPULDL · OmniScenes runtimetotal runtime (sec): 6.890 -> 0.738Improved

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