AI Paper Atlas

Research map / Point clouds / Segmentation and detection

GOOD: Geometry-guided Out-of-Distribution Modeling for Open-set Test-time Adaptation in Point Cloud Semantic Segmentation

Tianpei Zou, Guo Yu, Ya Wu, Fan Lu, Zhongcong Xu, Zhang Bo, Ziqiao Wang, Sanqing Qu, Guang Chen

ICLR 2026 · Point clouds › Segmentation and detection · approach: Point cloud semantic segmentation

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 Superpoint Confidence to temporal pseudo-labels (ablation)GOOD · Synth4D to SemanticKITTIAUROC: 66.66 -> 71.47Improved
add Superpoint Confidence to temporal pseudo-labels (ablation)GOOD · Synth4D to SemanticKITTImIoU gain: +4.61 -> +4.54Hurt
add Superpoint ID Prototypes (ablation)GOOD · Synth4D to SemanticKITTIAUROC / FPR95: AUROC 71.47 -> 73.39; FPR95 76.82 -> 71.91Improved
add Superpoint ID Prototypes (ablation)GOOD · Synth4D to SemanticKITTImIoU gain: +4.54 -> +4.83Improved
add temporal pseudo-labels alone (ablation)GOOD · Synth4D to SemanticKITTImIoU gain: +2.90 -> +4.61Improved
add GOOD to HGLHGL · Synth4D to SemanticKITTImIoU gain / FPR95: mIoU +2.90 -> +5.31; FPR95 79.82 -> 71.31Improved
replace HGL with GOODGOOD · SynLiDAR to SemanticKITTImIoU gain / AUROC: +1.90 -> +5.03; 62.91 -> 73.23Improved
apply 2D UniEnt entropy maximization to 3D TTAGIPSO · Synth4D to SemanticKITTImIoU gain: +2.08 -> +1.17Hurt
add WCEM to HGLHGL · Synth4D to SemanticKITTIFPR95: 79.82 -> 100.0No clear effect
add multi-frame superpoint clustering (ablation)GOOD · Synth4D to SemanticKITTIruntime: 0.70 s -> 1.45 sHurt
increase lambda for OOD entropy loss (ablation)GOOD · Synth4D to SemanticKITTImIoU / open-setMixed
use SRB under sparse unknown class (other-vehicle) (ablation)GOOD · SynLiDAR to SemanticKITTIAUROC: 65.35 -> 73.23Improved

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