Research map / Image generation / Autoregressive and latent generation
ACPV-Net: All-Class Polygonal Vectorization for Seamless Vector Map Generation from Aerial Imagery
Weiqin Jiao, Hao Cheng, George Vosselman, Claudio Persello
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 ACPV-Net's single-run all-class output instead of per-class baselines | GCP · Deventer-512 Vegetation | intra-class overlap (%): 29.05 -> 0.00 | Improved |
| use ACPV-Net instead of HiSup (CIoU) | HiSup · Shanghai | CIoU: 52.8 -> 64.9 | Improved |
| use ACPV-Net instead of HiSup (AP) | HiSup · Shanghai | AP: 45.0 -> 42.1 | Hurt |
| use ACPV-Net instead of HiSup (PoLiS) | HiSup · Shanghai | PoLiS: 2.4 -> 2.2 | Improved |
| use ACPV-Net instead of TopDiG | TopDiG · Deventer-512 | F1 topo: 49.6 -> 74.9 | Improved |
| use ACPV-Net instead of TopDiG | TopDiG · Deventer-512 | APLS: 50.7 -> 76.2 | Improved |
| use latent distributional vertex modeling instead of pure discriminative decoder (ablation) | ACPV-Net · Deventer-512 NMS threshold sweep | IoU, PoLiS, N-ratio | Improved |
| use vertex-guided subset selection instead of Douglas-Peucker simplification (ablation) | PSLG simplification · Deventer-512 | PoLiS vs redundancy | Improved |
| use cross-attention instead of channel concatenation for semantic conditioning (ablation) | ACPV-Net SSC · Deventer-512 | quality | No clear effect |
| use a deeper denoising UNet (ablation) | ACPV-Net · Deventer-512 | quality | No clear effect |
Closest papers by meaning
- Patched Line Segment Learning for Vector Road Mapping (AAAI 2024 · similarity 0.86)
- Joint Semantic-geometric Learning for Polygonal Building Segmentation (AAAI 2021 · similarity 0.86)
- SkyScenes: A Synthetic Dataset for Aerial Scene Understanding (ECCV 2024 · similarity 0.85)
- Topological Map Extraction From Overhead Images (ICCV 2019 · similarity 0.85)
- Predicting Ground-Level Scene Layout From Aerial Imagery (CVPR 2017 · similarity 0.85)
- SatSynth: Augmenting Image-Mask Pairs through Diffusion Models for Aerial Semantic Segmentation (CVPR 2024 · similarity 0.85)
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.
