AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

DOGE: Differentiable Bézier Graph Optimization for Road Network Extraction

Jiahui Sun, Junran Lu, Jinhui Yin, Yishuo Xu, Yuanqi Li, Yanwen Guo

ECCV 2026 · Graph learning › Graph algorithms and clustering · approach: Graph embedding and spectral sparsification · 2 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 TopoAdapt topology refinement (ablation)DOGE · SpaceNetAPLS: 69.45 -> 73.48Improved
add TopoAdapt topology refinement (ablation)DOGE · SpaceNetTOPO F1: 82.80 -> 84.58Improved
add overlap loss (ablation)DOGE · SpaceNetAPLS: 65.95 -> 73.48Improved
add offset and spacing curve regularization (ablation)DOGE · SpaceNetAPLS: 65.97 -> 73.48Improved
add all geometric priors (ablation)DOGE · SpaceNetTOPO F1: 74.58 -> 84.58Improved
add geometric priors and TopoAdapt to coverage-loss-only baseline (ablation)DOGE · SpaceNetAPLS: 37.10 -> 73.48Improved
add geometric priors only (without TopoAdapt) to coverage-loss baseline (ablation)L2-only baseline · SpaceNetTOPO F1: 70.94 -> 82.80Improved
use Bezier graph with global optimization instead of SAMRoad++ polylinesSAMRoad++ · SpaceNetTOPO F1: 81.57 -> 84.58Improved
use DOGE instead of SAMRoad++SAMRoad++ · City-ScaleAPLS: 68.34 -> 70.24Improved
use DOGE instead of SAMRoad++SAMRoad++ · City-ScaleTOPO Precision: 88.39 -> 84.42Hurt
use curve-based Bezier graph instead of polylinesSAMRoad++ · City-ScaleEdges/km: 44.49 -> 14.00Improved

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