Research map / Label-efficient and robust learning / Self-supervised and contrastive
Holistically-Attracted Wireframe Parsing: From Supervised to Self-Supervised Learning
Nan Xue, Tianfu Wu, Song Bai, Fudong Wang, Gui-Song Xia, Liangpei Zhang, Philip H. S. Torr
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 |
|---|---|---|---|
| upgrade HAWPv1 to HAWPv2 | HAWP · Wireframe | sAP5: 62.5 -> 65.7 | Improved |
| upgrade HAWPv1 to HAWPv2 | HAWP · YorkUrban | sAP5: 26.1 -> 28.8 | Improved |
| upgrade HAWPv1 to HAWPv2 | HAWP · Wireframe | FPS: 29.5 -> 40.8 | Improved |
| add end-point decoupled feature aggregation (ablation) | LOIAlign · Wireframe | sAP5: 64.0 -> 65.3 | Improved |
| add features of intermediate points on proposed line segments (ablation) | EPD LOIAlign · Wireframe | sAP5: 65.3 -> 65.4 | No clear effect |
| add auxiliary score BCE loss (ablation) | EPD LOIAlign · Wireframe | sAP5: 65.4 -> 65.7 | Improved |
| add EPE loss and binding threshold 10 at test (ablation) | HAWPv2 proposals · Wireframe | # proposals: 6.19k -> 2.24k | Improved |
| add EPE loss and binding threshold 10 at test (ablation) | HAWPv2 · Wireframe | sAP5: 65.5 -> 65.7 | No clear effect |
| add distance residual learning (ablation) | HAWPv2 · Wireframe | sAP5: 61.1 -> 62.9 | Improved |
| add rectified distance maps with residual scales -2 to +2 (ablation) | HAWPv2 · Wireframe | sAP5: 62.9 -> 65.7 | Improved |
| use unsigned instead of signed residuals (ablation) | HAWPv2 · Wireframe | sAP5: 62.6 -> 62.9 | Improved |
| use HAWPv2 instead of AFM++ on heatmap F score | HAWPv2 · Wireframe | F^H: 82.8 -> 81.4 | Hurt |
Closest papers by meaning
- Holistically-Attracted Wireframe Parsing (CVPR 2020 · similarity 0.94)
- End-to-End Wireframe Parsing (ICCV 2019 · similarity 0.92)
- NEAT: Distilling 3D Wireframes from Neural Attraction Fields (CVPR 2024 · similarity 0.86)
- Spot the Error: Non-autoregressive Graphic Layout Generation with Wireframe Locator (AAAI 2024 · similarity 0.81)
- COCOTree: A Dataset and Benchmark for Open Tree-Structured Visual Decomposition (NeurIPS 2026 · similarity 0.80)
- SegGraph: Leveraging Graphs of SAM Segments for Few-Shot 3D Part Segmentation (NeurIPS 2025 · similarity 0.80)
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.
