Research map / Graph learning / Knowledge and temporal graphs
DAGMapper: Learning to Map by Discovering Lane Topology
Namdar Homayounfar, Wei-Chiu Ma, Justin Liang, Xinyu Wu, Jack Fan, Raquel Urtasun
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add global feature network (ablation) | local-ROI DAGMapper · Highway 1 test (3-fold CV) | F1 @2px: 68.1 -> 69.5 | Improved |
| add Convolutional RNN memory to global feature network (ablation) | DAGMapper (G+R) · Highway 1 test (3-fold CV) | F1 @2px: 73.1 -> 76.3 | Improved |
| add Convolutional RNN memory to global feature network (ablation) | DAGMapper (G only) · Highway 1 test (3-fold CV) | F1 @2px: 69.5 -> 76.0 | Improved |
| add DT-prediction re-initialization (recovery) (ablation) | DAGMapper (G+M) · Highway 1 test (3-fold CV) | F1 @2px: 76.0 -> 76.3 | Improved |
| add DT-prediction recovery (ablation) | DAGMapper (G+M) · Highway 1 test (3-fold CV) | precision @2px: 77.0 -> 76.4 | Hurt |
| add DT recovery to global feature network (ablation) | DAGMapper (G only) · Highway 1 test (3-fold CV) | F1 @2px: 69.5 -> 73.1 | Improved |
| add notion of state to the nodes (ablation) | DAGMapper (G+M+R) · Highway 1 test (3-fold CV) | F1 @2px: 70.0 -> 76.3 | Improved |
| add notion of state to the nodes (ablation) | DAGMapper (G+M+R) · Highway 1 test (3-fold CV) | F1 @10px: 93.8 -> 96.3 | Improved |
| DAGMapper versus distance-transform baseline at tight threshold | DT baseline · Highway 1 test (3-fold CV) | F1 @2px: 66.9 -> 76.3 | Improved |
| DAGMapper versus distance-transform baseline at loose threshold | DT baseline · Highway 1 test (3-fold CV) | F1 @10px: 97.0 -> 96.3 | Hurt |
| DAGMapper versus HRAN with GT initialization | HRAN (GT init) · Highway 1 test (3-fold CV) | F1 @2px: 50.1 -> 76.3 | Improved |
| test on a different highway in another region without adaptation | DAGMapper · Highway generalization | F1 @2px: 76.3 -> 55.1 | Hurt |
Closest papers by meaning
- LaneSegNet: Map Learning with Lane Segment Perception for Autonomous Driving (ICLR 2024 · similarity 0.91)
- T2SG: Traffic Topology Scene Graph for Topology Reasoning in Autonomous Driving (CVPR 2025 · similarity 0.91)
- TopoMLP: A Simple yet Strong Pipeline for Driving Topology Reasoning (ICLR 2024 · similarity 0.90)
- LaneGraph2Seq: Lane Topology Extraction with Language Model via Vertex-Edge Encoding and Connectivity Enhancement (AAAI 2024 · similarity 0.90)
- TopoLogic: An Interpretable Pipeline for Lane Topology Reasoning on Driving Scenes (NeurIPS 2024 · similarity 0.90)
- TopoPoint: Enhance Topology Reasoning via Endpoint Detection in Autonomous Driving (NeurIPS 2025 · similarity 0.90)
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.
