AI Paper Atlas

Research map / 3D vision / Face and shape

Neural Vector Fields for Implicit Surface Representation and Inference

Edoardo Mello Rella, Ajad Chhatkuli, Ender Konukoğlu, Luc Van Gool

IJCV 2024 · 3D vision › Face and shape · approach: Learned mesh and geometric representations · 14 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
predict vector field directly instead of unsigned distance (NDF)Neural Vector Fields · ShapeNet watertight chairsChamfer mean: 0.418 -> 0.193Improved
predict vector field directly instead of NDFNeural Vector Fields · MGN clothes (open)Chamfer mean: 0.786 -> 0.128Improved
use VF instead of SDF-based DeepSDFNeural Vector Fields · ShapeNet watertight sofasChamfer mean: 0.120 -> 0.152Hurt
use VF instead of DeepSDFNeural Vector Fields · ShapeNet watertight chairsF1-Score: 89.31 -> 87.57Hurt
apply marching cubes on NDF distance instead of directly predicting directionsNDF · ShapeNet watertight chairsChamfer mean: 0.418 -> 0.629Hurt
add piecewise planar prior (Planar VF) (ablation)VF · ShapeNet planar bookshelvesChamfer mean: 1.116 -> 0.558Improved
add piecewise planar prior (Planar VF) (ablation)VF · ShapeNet planar laptopsF1-Score: 94.38 -> 95.76Improved
use larger backbone network (ablation)VF · ShapeNet cars (larger backbone)Chamfer mean: 0.0122 vs NDF 0.0126Improved
compare VF to differentiated UDF for normalsVF vs UDF · Normal consistency (cube toy)L1 distance: 0.149 -> 0.125Improved
use VF instead of UDF gradients on tetrahedron shapeVF vs UDF · Double tetrahedronL1 error: 0.4862 -> 0.1831Improved
use VF instead of UDF gradients on stress testVF vs UDF · Complex tetrahedron stress testL1 error: 0.7952 -> 0.3085Improved
use VF instead of SDF/UDF for normal consistencyVF · ShapeNet chairs normalscosine similarity: 0.874 -> 0.898Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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