AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

Learn-to-Score: Efficient 3D Scene Exploration by Predicting View Utility

Benjamin Hepp, Debadeepta Dey, Sudipta N. Sinha, Ashish Kapoor, Neel Joshi, Otmar Hilliges

ECCV 2018 · 3D vision › NeRF and radiance fields · approach: Inverse rendering with neural fields · 58 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
use learned 3D ConvNet utility instead of ProximityCountnext-best-view exploration · Washington2efficiency (normalized): 0.52 -> 0.91Improved
use learned utility instead of the best handcrafted utilitynext-best-view exploration · Neighborhoodefficiency (normalized): 0.60 -> 0.74Improved
use learned utility instead of Frontiernext-best-view exploration · Parisefficiency (normalized): 0.57 -> 0.87Improved
use learned utility instead of Frontiernext-best-view exploration · SanFranciscoefficiency (normalized): 0.09 -> 0.77Improved
use learned utility instead of ProximityCount, trained on one scenenext-best-view exploration · Washington1efficiency (normalized): 0.47 -> 0.88Improved
replace raycast-based AverageEntropy/ProximityCount scoring with a ConvNet forward passutility function · per-step computationtime per step (s): 8.35 -> 0.57Improved
replace Frontier raycast with learned utilityutility function · per-step computationtime per step (s): 0.61 -> 0.57No clear effect
apply low sensor noise at test time (40% pixels dropped, sigma 0.1m)Learn-to-Score · Washington2efficiency (normalized): 1.00 -> 0.99No clear effect
apply very high sensor noise at test time (sigma 1.0m)Learn-to-Score · Washington2efficiency (normalized): 1.00 -> 1.02No clear effect
use stereo-matched depth images from a photorealistic engineLearn-to-Score · Washington2efficiency (normalized): 1.00 -> 0.99No clear effect
use learned utility trained on one real scene instead of ProximityCountnext-best-view exploration · outdoor real buildingefficiency (normalized): 0.58 -> 0.90Improved
use learned utility trained on Area1 instead of ProximityCountnext-best-view exploration · indoor 2D-3D-S Area5befficiency (normalized): 0.52 -> 0.78Improved

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