Research map / Detection and segmentation / CNN features
DeepDualMapper: A Gated Fusion Network for Automatic Map Extraction Using Aerial Images and Trajectories
Hao Wu, Hanyuan Zhang, Xinyu Zhang, Weiwei Sun, Baihua Zheng, Yuning Jiang
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 |
|---|---|---|---|
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Porto | IoU: 0.688 \pm 0.021 -> 0.717 \pm 0.003 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Shanghai | IoU: 0.626 \pm 0.019 -> 0.634 \pm 0.012 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Singapore | IoU: 0.605 \pm 0.029 -> 0.620 \pm 0.025 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Porto | F1-score: 0.815 \pm 0.015 -> 0.835 \pm 0.002 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Shanghai | F1-score: 0.770 \pm 0.015 -> 0.776 \pm 0.009 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of V-FuseNet | Map extraction models · Singapore | F1-score: 0.754 \pm 0.023 -> 0.765 \pm 0.019 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of the strongest attack baseline | Map extraction models · Porto (information loss attack) | IoU: 0.546\pm0.025 -> 0.588\pm0.019 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of the strongest attack baseline | Map extraction models · Shanghai (information loss attack) | IoU: 0.467\pm0.022 -> 0.497\pm0.023 | Improved |
| Use gated multimodal fusion and densely supervised refinement instead of the strongest attack baseline | Map extraction models · Singapore (information loss attack) | IoU: 0.445\pm0.016 -> 0.477\pm0.030 | Improved |
| Add the gated fusion step to combine image and trajectory features (ablation) | DeepDualMapper · Map extraction (Porto; figure-described study) | IoU | Improved |
| Add refinement decoding after gated fusion (ablation) | DeepDualMapper · Map extraction (Porto; figure-described study) | IoU | Improved |
| Add dense supervision to the fused feature (ablation) | DeepDualMapper · Map extraction (Porto; figure-described study) | IoU | Improved |
Closest papers by meaning
- DeepRoadMapper: Extracting Road Topology From Aerial Images (ICCV 2017 · similarity 0.87)
- Leveraging Crowdsourced GPS Data for Road Extraction From Aerial Imagery (CVPR 2019 · similarity 0.86)
- Learning to Generate Maps from Trajectories (AAAI 2020 · similarity 0.85)
- Patched Line Segment Learning for Vector Road Mapping (AAAI 2024 · similarity 0.82)
- Multi-Modal Aerial-Ground Cross-View Place Recognition with Neural ODEs (CVPR 2025 · similarity 0.82)
- DeepPointMap: Advancing LiDAR SLAM with Unified Neural Descriptors (AAAI 2024 · similarity 0.82)
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.
