AI Paper Atlas

Research map / 3D vision / Face and shape

PartCom: Part Composition Learning for 3D Open-Set Recognition

Tingyu Weng, Jun Xiao, Hao Pan, Haiyong Jiang

IJCV 2023 · 3D vision › Face and shape · approach: Primitive-based shape abstraction · 7 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 part prototypes without constraints (ablation)GCP baseline · ModelNet 20-20OSCR: 85.5 -> 86.6Improved
add part diversity loss L_pd to part prototypes (ablation)part prototype model · ModelNet 20-20OSCR: 86.6 -> 93.9Improved
add part balance loss L_pb to part prototypes (ablation)part prototype model · ModelNet 20-20OSCR: 86.6 -> 93.6Improved
add both L_pd and L_pb (ablation)part prototype model · ModelNet 20-20OSCR: 86.6 -> 95.8Improved
replace feature concatenation with max pooling (ablation)PartCom · ModelNet 20-20OSCR: 96.7 -> 94.4Hurt
add PUFS unknown feature synthesis (ablation)PartCom without PUFS · ModelNet 20-20OSCR: 95.8 -> 96.7Improved
increase number of part prototypes per class from 1 to 3 or more (ablation)PartCom · ModelNet 20-20OSCRImproved
increase cosine similarity threshold delta above 0.1 (ablation)PartCom · ModelNet 20-20OSCRHurt
replace ARPL with PartCom on confusing mixup tasks3D OSR with PointNet++ · ShapeNetMixOSCR: 58.4 -> 99.1Improved
use PartCom instead of ARPL3D OSR with PointNet++ · ScanObjectNN 5-10OSCR: 84.7 -> 97.1Improved
use class-specific prototype methods (GCP, RPL) instead of softmax3D closed-set classifier · ModelNetMixOSCR: 80.0 -> 63.8Hurt

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