AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Learning Where to Look: Self-supervised Viewpoint Selection for Active Localization using Geometrical Information

Luca Di Giammarino, Boyang Sun, Giorgio Grisetti, Marc Pollefeys, Hermann Blum, Daniel Barath

ECCV 2024 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 1 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
learn viewpoint score with an MLP on binned landmark reprojections instead of Fisher information fieldactive viewpoint selection · Habitat-Matterport 3D test, 5m 10deglocalization score (%): 69.6 -> 81.7Improved
choose viewpoints by uniform binned feature reprojection (no learning) vs Fisher informationbaseline · Habitat-Matterport 3D test, 0.25m 2deglocalization score (%): 67.7 -> 71.7Improved
add learned MLP on top of reprojection-with-bins heuristicviewpoint selection · Habitat-Matterport 3D test, 5m 10deglocalization score (%): 71.9 -> 81.7Improved
add depth (z) to 2D landmark projections as input (ablation)viewpoint MLP · Habitat-Matterport 3D test, 5m 10deglocalization score (%): 76.2 -> 80.6Improved
use full 3D landmarks instead of depth only (ablation)viewpoint MLP · Habitat-Matterport 3D test, 0.05m 0.4deglocalization score (%): 74.6 -> 72.9Hurt
use full 3D landmarks instead of depth only (ablation)viewpoint MLP · Habitat-Matterport 3D test, 5m 10deglocalization score (%): 80.6 -> 81.7Improved
use GPU CUDA implementation for viewpoint samplingsampling pipeline · 648 locations, 24919 landmarksruntime: 7387 s -> 23 sImproved

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