Research map / Detection and segmentation / CNN features
SFD2: Semantic-Guided Feature Detection and Description
Fei Xue, Ignas Budvytis, Roberto Cipolla
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 semantic detection loss (ablation) | SFD2 · Aachen v1.1 Night | localization success ratio at (2°,0.25m): 71.2 -> 72.8 | Improved |
| Add semantic description loss with triplet intra-class loss (ablation) | SFD2 · Aachen v1.1 Night | localization success ratio at (2°,0.25m): 71.2 -> 73.8 | Improved |
| Use ranking instead of triplet intra-class loss in semantic description (ablation) | SFD2 · Aachen v1.1 Night | localization success ratio at (2°,0.25m): 73.8 -> 74.9 | Improved |
| Add semantic feature-consistency loss to SD + SS (ablation) | SFD2 · Aachen v1.1 Night | localization success ratio at (2°,0.25m): 75.9 -> 78.0 | Improved |
| Add semantic description loss to semantic detection (ablation) | SFD2 · Aachen v1.1 Day | localization success ratio at (2°,0.25m): 87.3 -> 88.2 | Improved |
| Add semantic description loss to semantic detection (ablation) | SFD2 · Aachen v1.1 Night | localization success ratio at (2°,0.25m): 72.8 -> 75.9 | Improved |
| Use SFD2 instead of R2D2 | local feature extraction · RobotCar-Seasons night | localization success ratio at (2°,0.25m): 18.3 -> 27.6 | Improved |
| Use SFD2 instead of R2D2 | feature extraction · 1024×1024 input | running time (ms): 72.4 -> 33.2 | 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.
- R2D2 — R2D2: Reliable and Repeatable Detector and Descriptor (NeurIPS 2019) · better on 1 of 1 shared result
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
- Semantic Visual Localization (CVPR 2018 · similarity 0.89)
- Semantic Match Consistency for Long-Term Visual Localization (ECCV 2018 · similarity 0.89)
- SemanticKITTI: A Dataset for Semantic Scene Understanding of LiDAR Sequences (ICCV 2019 · similarity 0.87)
- SemLight: Distilled Semantic–Geometric Fusion for Efficient Local Feature Matching (ECCV 2026 · similarity 0.87)
- Recognize Any Regions (NeurIPS 2024 · similarity 0.86)
- Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models (NeurIPS 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.
