AI Paper Atlas

Research map / 3D vision / Face and shape

Semantic-Aware Implicit Template Learning via Part Deformation Consistency

Sihyeon Kim, Minseok Joo, Jaewon Lee, Juyeon Ko, Juhan Cha, Hyunwoo J. Kim

ICCV 2023 · 3D vision › Face and shape · approach: Primitive-based shape abstraction · 5 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 soft instead of hard assignment of part deformation priors (ablation)semantic-aware deformation code · ShapeNet chair 5-shot part label transfermIoU: 77.0 -> 84.2Improved
add global deformation consistency (Chamfer) to part deformation consistency (ablation)regularization · ShapeNet chair 5-shot PLTmIoU: 83.4 -> 83.9Improved
add global scale consistency regularization (ablation)regularization · ShapeNet chair 5-shot PLTmIoU: 83.9 -> 84.2Improved
use only part deformation consistency losses (ablation)regularization · ShapeNet chair 5-shot PLTmIoU: 80.4/82.1 (DIT/DIF) -> 83.4Improved
vary BAE-Net part count k from 2 to 10 (ablation)semantic prior · ShapeNet chair 5-shot PLTmIoU: 82.8 to 83.8 across kNo clear effect
semantic-aware deformation and part consistency on high-variability categoriesproposed framework vs DIF · KeypointNet chair PCK@0.1PCK: 57.6 -> 83.2Improved
semantic prior vs DIF in reconstruction of unseen table shapesproposed framework · ShapeNet table reconstructionChamfer distance x10^3: 0.2640 -> 0.2796Hurt
semantic prior in 2-shot part label transfer across subcategoriesproposed framework vs DIF · ShapeNet subcategoriesmIoU: 64.5 -> 72.8 (airliner/jet); 66.3 -> 77.5 (straight/sofa)Improved
use coarser semantic prior (RIM-Net level 1) instead of finer one (ablation)semantic prior · ShapeNet chair subcategories 2-shot PLTmIoU: 78.3 -> 70.1Hurt
apply method to non-rigid DFAUSTproposed framework vs DIF · DFAUST keypoint transferPCK@0.05 / 0.1: 67.0/80.7 -> 74.9/85.2Improved

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