AI Paper Atlas

Research map / Point clouds / Shape and surface completion

Learning Local Pattern Modularization for Point Cloud Reconstruction from Unseen Classes

Chao Chen, Yu-Shen Liu, Zhizhong Han

ECCV 2024 · Point clouds › Shape and surface completion · approach: Unsupervised shape correspondence · 2 citations (OpenAlex)

Read the paper

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 local prior (pattern modularization plus customization) over initial shape decoder only (ablation)Ours · ShapeNet Bench (seen, 2048 pts)L1-CD: 0.054 -> 0.032Improved
add pattern learner and pattern modularization (ablation)Ours · ShapeNet BenchL1-CD: 0.049 -> 0.032Improved
add modularization customizer shift from the image (ablation)Ours · ShapeNet BenchL1-CD: 0.041 -> 0.032Improved
add region loss L_Region instead of repeating L_Shape (ablation)Ours · ShapeNet BenchL1-CD: 0.038 -> 0.032Improved
add weak global shape loss L_Shape (ablation)Ours · ShapeNet BenchL1-CD: 0.048 -> 0.032Improved
sample pattern points in 3D voxel instead of on a 2D plane (ablation)pattern learner · ShapeNet BenchL1-CD: 0.034 -> 0.032Improved
vary region number M from 8 to 1 or 27 (ablation)Ours · ShapeNet BenchL1-CD: 0.032 -> 0.035 (M=1), 0.049 (M=27)Hurt
increase pattern number from 8 to 16 (ablation)Ours · ShapeNet BenchL1-CD: 0.032 -> 0.035Hurt
use local prior instead of global prior for unseen class reconstructionOurs vs DRWR · ShapeNet 4 unseen classes (1024 pts)L1-CD: 0.0715 -> 0.0540Improved
train on 3 seen classes while competitors use their full training settingsOurs · ShapeNet seen (5-class protocol)L1-CD: 0.0536 -> 0.0527Improved
use ours versus SDT on Chair, Sofa and Table classesOurs · ShapeNet seen classes (2048 pts)L1-CD: Chair 0.036 -> 0.040; Table 0.033 -> 0.044Mixed

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