AI Paper Atlas

Research map / 3D vision / Face and shape

Geometry Processing with Neural Fields

Guandao Yang, Serge Belongie, Bharath Hariharan, Vladlen Koltun

NeurIPS 2021 · 3D vision › Face and shape · approach: Learned mesh and geometric representations

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
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · DinoCDr: 1.54 -> 1.04Improved
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · ArmadilloCDr: 1.36 -> 1.02Improved
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · DinoEMDr: 3.38 -> 1.15Improved
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · ArmadilloEMDr: 3.30 -> 1.08Improved
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · Dinosampling time: 0.15 -> 0.21Hurt
initialize Langevin dynamics from rejection-sampled rough shapeneural field surface sampling · Armadillosampling time: 0.12 -> 0.18Hurt
replace invertible deformation network with SIREN (no inverse) (ablation)neural field surface sampling · 2D rectangle SDF deformationtopology preservation (qualitative)Hurt
remove positional encoding from deformation network (No PE) (ablation)neural field surface sampling · 2D rectangle SDF deformationdeformation complexity (qualitative)Hurt
optimize only stretching loss Ls instead of both losses (ablation)neural field surface sampling · 2D rectangle SDF deformationshape quality (qualitative)Mixed
optimize only bending loss Lb instead of both losses (ablation)neural field surface sampling · 2D rectangle SDF deformationshape quality (qualitative)Mixed
apply ARAP to marching-cubes mesh instead of neural field thin-shell optimizationneural field surface sampling · Cactus, Armadillo, Dino deformationvolume distortion (qualitative)Hurt
remesh (simplify) extracted mesh before ARAP, or use SR-ARAPneural field surface sampling · ARAP baselinesdeformation quality (qualitative)Improved

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