AI Paper Atlas

Research map / Point clouds / Shape and surface completion

GeoUDF: Surface Reconstruction from 3D Point Clouds via Geometry-guided Distance Representation

Siyu Ren, Junhui Hou, Xiaodong Chen, Ying He, Wenping Wang

ICCV 2023 · Point clouds › Shape and surface completion · approach: Distance-function surface reconstruction · 67 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 LGR local geometry upsampling before reconstruction (ablation)GeoUDF · ShapeNet (clean)Chamfer distance mean (x10^-2): 0.284 -> 0.234Improved
add LGR upsampling to GIFS (ablation)GIFS · ShapeNet (clean)Chamfer distance mean (x10^-2): 0.328 -> 0.316Improved
use point-to-tangent-plane distance instead of point-to-point (ablation)GUE · ShapeNet UDF accuracyUDF error (x10^-3): 1.240 -> 0.615Improved
add neighborhood shape encoding (ablation)GUE · ShapeNet UDF accuracyUDF error (x10^-3): 1.453 -> 0.615Improved
estimate gradient geometrically instead of back-propagation (ablation)GUE · ShapeNet UDF accuracytime (us/point): 12.125 -> 3.405Improved
decouple UDF and gradient estimation versus regressing both with MLPs (ablation)GUE · ShapeNet UDF accuracygradient error (degrees): 8.371 -> 7.237Improved
use E-MC instead of MeshUDF for mesh extraction (ablation)GeoUDF · ShapeNet (clean)Chamfer distance mean (x10^-2): 0.266 -> 0.234Improved
vary neighborhood size K from 5 to 20 (ablation)GeoUDF · ShapeNet (clean)Chamfer distance mean (x10^-2): 0.235 -> 0.232No clear effect
increase E-MC grid resolution from 128 to 192 (ablation)GeoUDF · ShapeNet (clean)Chamfer distance mean (x10^-2): 0.234 -> 0.223Improved
use GeoUDF instead of GIFS on unseen non-watertight shapesGeoUDF · MGNChamfer distance mean (x10^-2): 0.248 -> 0.194Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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