AI Paper Atlas

Research map / Point clouds / Segmentation and detection

Query Refinement Transformer for 3D Instance Segmentation

Jiahao Lu, Jiacheng Deng, Chuxin Wang, Jianfeng He, Tianzhu Zhang

ICCV 2023 · Point clouds › Segmentation and detection · approach: Scene completion and instance segmentation · 37 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 Query Initialization Module (ablation)Mask3D baseline · ScanNetV2 valmAP: 53.7 -> 55.3Improved
add cluster-based layer (ablation)Mask3D baseline · ScanNetV2 valmAP: 53.7 -> 54.0No clear effect
add denoising module (ablation)QIM + cluster-based layer · ScanNetV2 valmAP: 55.4 -> 56.2Improved
add DBSCAN post-processing (ablation)QueryFormer · ScanNetV2 valmAP: 56.2 -> 56.5No clear effect
add DBSCAN post-processingMask3D · S3DIS Area 5mAP: 53.9 -> 56.5Improved
increase number of queries from 100 to 500 (ablation)Mask3D · ScanNetV2 valmAP: 53.7 -> 52.8Hurt
increase number of queries from 100 to 150 (ablation)QueryFormer · ScanNetV2 valmAP: 56.2 -> 55.3Hurt
replace FPS query sampling with QIM at 100 queriesMask3D · ScanNetV2 valcoverage rate / repetition rate: CR 0.74 -> 0.89; RR 0.64 -> 0.47Improved
use cluster-based layer instead of the normal decoder layersingle decoder layer · ScanNetV2 valFLOPs (G): 2.3 -> 1.3Improved

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