Research map / Detection and segmentation / CNN features
DeepLSD: Line Segment Detection and Refinement With Deep Image Gradients
Read the paperCheck 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 |
|---|---|---|---|
| Use DeepLSD instead of LSD | Line detection · RDNIM | homography estimation: 0.468 -> 0.591 | Helped |
| Use DeepLSD instead of LSD | Line detection · HPatches | homography estimation: 0.697 -> 0.705 | Helped |
| Use DeepLSD instead of LSD | Line detection · Hypersim average | 3D line recall: 213.9 -> 226.1 | Helped |
| Use DeepLSD instead of LSD | 7Scenes localization · 7Scenes Total | pose accuracy at 5 cm / 5 deg: 83.4 -> 84.0 | Helped |
| Detect double edges rather than single edges (ablation) | DeepLSD · HPatches | homography estimation: 0.434 -> 0.705 | Helped |
| Add distance-field normalization (ablation) | DeepLSD · HPatches | homography estimation: 0.674 -> 0.705 | Helped |
| Replace hybrid LSD extraction with HAWP trained on the pseudo-ground-truth lines (ablation) | DeepLSD · HPatches | homography estimation: 0.705 -> 0.245 | Hurt |
| Add line refinement without vanishing-point constraints | HAWP · Wireframe | homography estimation: 0.701 -> 0.864 | Helped |
| Add line refinement with vanishing-point constraints | TP-LSD · Wireframe | homography estimation: 0.658 -> 0.913 | Helped |
| Add vanishing-point constraints to line refinement (ablation) | HAWP · Wireframe | homography estimation: 0.864 -> 0.892 | Helped |
| Add vanishing-point constraints to line refinement (ablation) | SOLD2 · Wireframe | homography estimation: 0.935 -> 0.920 | Hurt |
| Add vanishing-point constraints to line refinement (ablation) | DeepLSD · Wireframe | homography estimation: 0.994 -> 0.981 | Hurt |
Closest papers by meaning
- ScaleLSD: Scalable Deep Line Segment Detection Streamlined (CVPR 2025 · similarity 0.92)
- Learning Attraction Field Representation for Robust Line Segment Detection (CVPR 2019 · similarity 0.90)
- SOLD2: Self-Supervised Occlusion-Aware Line Description and Detection (CVPR 2021 · similarity 0.87)
- Towards Light-Weight and Real-Time Line Segment Detection (AAAI 2022 · similarity 0.87)
- Improving Transformer Based Line Segment Detection with Matched Predicting and Re-ranking (AAAI 2025 · similarity 0.87)
- MonoDGP: Monocular 3D Object Detection with Decoupled-Query and Geometry-Error Priors (CVPR 2025 · similarity 0.84)
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.