AI Paper Atlas

Research map / Detection and segmentation / CNN features

PolyWorld: Polygonal Building Extraction With Graph Neural Networks in Satellite Images

CVPR 2022 · Detection and segmentation › CNN features · approach: Boundary-aware instance segmentation · 111 citations (OpenAlex)

Check your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 vertex position refinement offsets learned with the refinement losses (ablation)PolyWorld · CrowdAI testAP: 58.7 -> 63.3Helped
add vertex position refinement offsets learned with the refinement losses (ablation)PolyWorld · CrowdAI testIoU: 89.9 -> 91.3Helped
combine clockwise and counterclockwise connection scores (ablation)PolyWorld · CrowdAI testAP: 62.1 -> 63.3; 60.3 -> 63.3Helped
combine clockwise and counterclockwise connection scores (ablation)PolyWorld · CrowdAI testIoU: 90.9 -> 91.3; 88.4 -> 91.3Helped
add the attentional GNN for global descriptor aggregation and vertex refinement (ablation)PolyWorld · CrowdAI testAP: 56.8 -> 58.7Helped
add the angle loss for learning vertex offsets (ablation)PolyWorld · CrowdAI testMTA: 34.0° -> 32.9°Helped
add the angle loss for learning vertex offsets (ablation)PolyWorld · CrowdAI testAP: 63.6 -> 63.3Mixed

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