AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Fine-Grained Representation for Lane Topology Reasoning

Guoqing Xu, Yiheng Li, Yang Yang

AAAI 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Dense features and foundation priors · 1 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 Hierarchical Prior Extractor (ablation)TopoLogic baseline · OpenLane-V2 subset AOLS: 44.1 -> 45.4Improved
add Region-Focused Decoder (ablation)TopoLogic + HPE · OpenLane-V2 subset AOLS: 45.4 -> 45.8Improved
add Robust Boundary-Point Topology Reasoning (ablation)TopoLogic + HPE + RFD · OpenLane-V2 subset AOLS: 45.8 -> 48.0Improved
add RFD (ablation)TopoLogic + HPE · OpenLane-V2 subset ADETt: 47.5 -> 47.2Hurt
add local sequential prior only (ablation)TopoLogic baseline · OpenLane-V2 subset ADETt: 47.2 -> 45.5Hurt
add global spatial prior on top of local prior (ablation)HPE · OpenLane-V2 subset ADETt: 45.5 -> 47.5Improved
add fine-grained query init without sampled reference points (ablation)TopoLogic + HPE · OpenLane-V2 subset ATOPll: 26.1 -> 28.6Mixed
add sampled reference points to fine-grained queries (ablation)RFD · OpenLane-V2 subset AOLS: 45.6 -> 45.8Improved
add boundary-point topology reasoning (ablation)TopoLogic + HPE + RFD · OpenLane-V2 subset ATOPll: 26.8 -> 28.3Improved
add denoised topology reasoning alone (ablation)TopoLogic + HPE + RFD · OpenLane-V2 subset ATOPlt: 27.7 -> 30.7Improved
add denoising on top of boundary-point reasoning (ablation)RBTR · OpenLane-V2 subset AOLS: 46.6 -> 48.0Improved
use TopoFG vs TopoLogicTopoFG · OpenLane-V2 subset BOLS: 42.3 -> 45.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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