Research map / Detection and segmentation / CNN features
Deep Hough-Transform Line Priors
Yancong Lin, Silvia L. Pintea, Jan C. van Gemert
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 |
|---|---|---|---|
| add convolutions in the Hough domain (vs none) (ablation) | HT-IHT block · Wireframe subset | AP: 61.77 -> 66.46 | Improved |
| use Laplacian-initialized channel-wise 1D conv instead of standard 1D conv (ablation) | HT-IHT block · Wireframe subset | AP: 63.02 -> 66.19 | Improved |
| add two extra 1D convs for merging and reducing channels (ablation) | HT-IHT block · Wireframe subset | AP: 66.19 -> 66.46 | Improved |
| use 1D convs over offsets instead of 3x3 convs in Hough domain (ablation) | HT-IHT block · Wireframe subset | AP: 63.90 -> 66.46 | Improved |
| replace hourglass blocks with HT-IHT blocks in LCNN | LCNN · Wireframe test | structural AP10: 62.9 -> 64.2 | Improved |
| replace hourglass blocks with HT-IHT blocks in LCNN | LCNN · York Urban test | junction mAP: 30.4 -> 32.5 | Improved |
| replace hourglass blocks with HT-IHT blocks in HAWP | HAWP · Wireframe test | structural AP10: 66.5 -> 66.6 | No clear effect |
| replace hourglass blocks with HT-IHT blocks in HAWP | HAWP · York Urban test | structural AP10: 28.5 -> 27.4 | Hurt |
| add Hough priors (HT-LCNN 5.9M) when training data is reduced | LCNN · Wireframe subsets (fewer samples) | structural AP10, junction mAP | Improved |
| add global Hough line prior to local-only model | single-filter network · Line-Circle | AP | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MCMLSD — MCMLSD: A Dynamic Programming Approach to Line Segment Detection (CVPR 2017) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Deep Hough Transform for Semantic Line Detection (ECCV 2020 · similarity 0.92)
- Deep Vanishing Point Detection: Geometric Priors Make Dataset Variations Vanish (CVPR 2022 · similarity 0.88)
- DeepLSD: Line Segment Detection and Refinement With Deep Image Gradients (CVPR 2023 · similarity 0.87)
- Learning Attraction Field Representation for Robust Line Segment Detection (CVPR 2019 · similarity 0.87)
- MCMLSD: A Dynamic Programming Approach to Line Segment Detection (CVPR 2017 · similarity 0.86)
- Improving Transformer Based Line Segment Detection with Matched Predicting and Re-ranking (AAAI 2025 · 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.
