AI Paper Atlas

Research map / Point clouds / Point cloud classification

Superpoint Network for Point Cloud Oversegmentation

Le Hui, Jia Yuan, Mingmei Cheng, Jin Xie, Xiaoya Zhang, Jian Yang

ICCV 2021 · Point clouds › Point cloud classification · approach: Local feature aggregation networks · 50 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 feature-space weights to coordinate-only association (ablation)SPNet · S3DIS Area 5boundary recall: 78.35 -> 82.11Improved
add feature-space weights to coordinate-only association (ablation)SPNet · S3DIS Area 5boundary precision: 9.43 -> 13.14Improved
add feature-space weights to coordinate-only association (ablation)SPNet · S3DIS Area 5oracle overall accuracy: 95.10 -> 96.50Improved
add coordinate-space weights to feature-only association (ablation)SPNet · S3DIS Area 5boundary recall: 82.03 -> 82.11No clear effect
add coordinate-space weights to feature-only association (ablation)SPNet · S3DIS Area 5boundary precision: 12.01 -> 13.14Improved
vary k (nearest superpoints) beyond k=1 (ablation)SPNet · S3DIS Area 5boundary recallNo clear effect
vary lambda (compactness) and iterations v (ablation)SPNet · S3DIS Area 5boundary recallNo clear effect
use SPNet superpoints instead of SSP superpoints for segmentationsuperpoint-based segmentation · vKITTI 6-foldmIoU: 52.0 -> 57.0Improved
use SPNet superpoints instead of SSP superpoints for segmentationsuperpoint-based segmentation · S3DIS 6-foldmIoU: 68.4 -> 68.7Improved
use SPNet superpoints instead of SSP superpoints for segmentationsuperpoint-based segmentation · S3DIS 6-foldinference time (s): 9581 -> 2828Improved
end-to-end SPNet instead of SSP with graph-cut post-processingsuperpoint generation · S3DIS Area 5inference time (s): 2327 -> 646Improved
use label consistency loss (SPNet vs SSP)superpoint generation · vKITTI 6-fold (w/ RGB)boundary recall: 90.37 -> 94.15Improved

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