Research map / Detection and segmentation / Open-set and few-shot detection
Polygonal Building Extraction by Frame Field Learning
Check 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 |
|---|---|---|---|
| Add frame field learning to the segmentation network (ablation) | UResNet101 · CrowdAI validation | mean max tangent angle error: 51.9° -> 45.1° | Helped |
| Add frame field learning to the segmentation network (ablation) | UResNet101 · CrowdAI | AP: 62.4 -> 64.5 | Helped |
| Add frame field learning to the segmentation network (ablation) | U-Net16 · Inria OSM test | IoU: 76.9% -> 78.0% | Helped |
| Add frame field learning to the segmentation network (ablation) | UResNet101 · Inria polygonized test | mean max tangent angle error: 52.0° -> 28.1° | Helped |
| Use frame-field-guided polygonization instead of simple polygonization (ablation) | UResNet101 (with field) · CrowdAI validation | mean max tangent angle error: 45.1° -> 31.9° | Helped |
| Use frame-field-guided polygonization instead of simple polygonization (ablation) | UResNet101 (with field) · CrowdAI | AP: 61.7 -> 61.3 | Hurt |
| Use frame-field-guided polygonization instead of simple polygonization (ablation) | UResNet101 (with field) · CrowdAI | AR: 65.4 -> 65.0 | Hurt |
| Apply frame-field-guided polygonization to U-Net variant [9] probability maps instead of ASIP | U-Net variant [9] · CrowdAI validation | mean max tangent angle error: 44.0° -> 36.6° | Helped |
| Apply frame-field-guided polygonization to U-Net variant [9] probability maps instead of ASIP | U-Net variant [9] · CrowdAI | AR: 78.7 -> 73.2 | Hurt |
| Use frame-field-guided polygonization instead of ASIP | UResNet101 (with field) · CrowdAI validation | mean max tangent angle error: 38.3° -> 31.9° | Helped |
| Use frame-field-guided polygonization instead of ASIP | UResNet101 (with field) · CrowdAI | AP: 60.0 -> 61.3 | Helped |
| Use frame-field-guided polygonization instead of ASIP | UResNet101 (with field) · CrowdAI | AR: 64.0 -> 65.0 | Helped |
Closest papers by meaning
- Joint Semantic-geometric Learning for Polygonal Building Segmentation (AAAI 2021 · similarity 0.90)
- PolyWorld: Polygonal Building Extraction With Graph Neural Networks in Satellite Images (CVPR 2022 · similarity 0.88)
- Efficient Interactive Annotation of Segmentation Datasets With Polygon-RNN++ (CVPR 2018 · similarity 0.84)
- SatSynth: Augmenting Image-Mask Pairs through Diffusion Models for Aerial Semantic Segmentation (CVPR 2024 · similarity 0.82)
- Aerial Lifting: Neural Urban Semantic and Building Instance Lifting from Aerial Imagery (CVPR 2024 · similarity 0.82)
- 3D Building Reconstruction from Monocular Remote Sensing Images with Multi-level Supervisions (CVPR 2024 · similarity 0.81)
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.