GeoReFormer: Geometry-Aware Refinement for Lane Segment Detection and Topology Reasoning

Abstract

Accurate 3D lane segment detection and topology reasoning are critical for structured online map construction in autonomous driving. Recent transformer-based approaches formulate this task as query-based set prediction, yet largely inherit decoder designs originally developed for compact object detection. However, lane segments are continuous polylines embedded in directed graphs, and generic query initialization and unconstrained refinement do not explicitly encode this geometric and relational structure. We propose GeoReFormer (Geometry-aware Refinement Transformer), a unified query-based architecture that embeds geometry- and topology-aware inductive biases directly within the transformer decoder. GeoReFormer introduces data-driven geometric priors for structured query initialization, bounded coordinate-space refinement for stable polyline deformation, and per-query gated topology propagation to selectively integrate relational context. On the OpenLane-V2 benchmark, GeoReFormer achieves state-of-the-art performance with 34.5% mAP while improving topology consistency over strong transformer baselines, demonstrating the utility of explicit geometric and relational structure encoding.

Figures (6)

• Figure 1: Overview of GeoReFormer, a geometry-aware query-based transformer for joint 3D lane segment detection and topology estimation from multi-view cameras
• Figure 2: Architecture of GeoReFormer. Multi-view images are lifted to BEV features and decoded from fixed geometric prios obtained via Spatially-Stratified K-Medoids. The decoder integrates bounded polyline refinement and per-query gated topology aggregation (TopoFFN). Prediction heads produce lane geometry and relational topology.
• Figure 3: Qualitative comparison of lane detection on the OpenLane-V2 benchmark. From left to right: ground truth, GeoReFormer, LaneSegNet and TopoLogic. The dotted ellipses show missing features correctly captured by GeoReFormer.
• Figure 4: Comparison between per-query and global topology gating on the Lane Segment Detection task
• Figure 5: Sensitivity of GeoReFormer to the residual scaling factor $s$. Performance trends for mAP and TOP$_{lsls}$ across different scaling values.