Research map / Detection and segmentation / CNN features
RoadTracer: Automatic Extraction of Road Networks From Aerial Images
Read the paperCheck 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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use iterative CNN-guided graph construction instead of segmentation and graph post-processing | RoadTracer versus Segmentation (Ours) · Junction metric, 5% error rate, 15 test cities | F_correct: 0.40 -> 0.58 | Helped |
| Use iterative CNN-guided graph construction instead of segmentation and graph post-processing | RoadTracer versus Segmentation (Ours) · SP, 15 test cities | Correct shortest paths: 0.58 -> 0.72 | Helped |
| Train on static oracle-generated examples rather than dynamically updating examples with the CNN's own search (ablation) | RoadTracer training · Road network inference | inference performance | Hurt |
| Add synthetic noise to the partial graph in static training examples (ablation) | RoadTracer training · Road network inference | inference performance | No clear effect |
| Use the CNN's predicted actions to update the graph during training while using the oracle only for labels (ablation) | RoadTracer training · Road network inference | robustness to imperfect partial graphs | Helped |
Closest papers by meaning
- VecRoad: Point-Based Iterative Graph Exploration for Road Graphs Extraction (CVPR 2020 · similarity 0.90)
- Improved Road Connectivity by Joint Learning of Orientation and Segmentation (CVPR 2019 · similarity 0.89)
- Patched Line Segment Learning for Vector Road Mapping (AAAI 2024 · similarity 0.87)
- Bezier Everywhere All at Once: Learning Drivable Lanes as Bezier Graphs (CVPR 2024 · similarity 0.85)
- Sketch and Refine: Towards Fast and Accurate Lane Detection (AAAI 2024 · similarity 0.84)
- Towards Satellite Image Road Graph Extraction: A Global-Scale Dataset and A Novel Method (CVPR 2025 · similarity 0.83)
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.