Research map / Detection and segmentation / CNN features
Neural Ctrl-F: Segmentation-Free Query-By-String Word Spotting in Handwritten Manuscript Collections
Tomas Wilkinson, Jonas Lindstrom, Anders Brun
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 DCToW embedding instead of PHOC (ablation) | Ctrl-F-Net · GW 5-15, 50% overlap, QbS | MAP (%): 65.6 -> 73.8 | Improved |
| use DCToW embedding instead of PHOC (ablation) | Ctrl-F-Net · GW 15-5, 50% overlap, QbE | MAP (%): 90.9 -> 90.5 | No clear effect |
| use DCToW embedding instead of PHOC (ablation) | Ctrl-F-Net · IAM, 50% overlap, QbS | MAP (%): 78.8 -> 80.3 | Improved |
| lower overlap threshold from 50% to 25% (ablation) | Ctrl-F-Net DCToW · GW 15-5, QbE | MAP (%): 90.5 -> 97.0 | Improved |
| add DTP proposals to RPN proposals (ablation) | Ctrl-F-Net · IAM, 50% overlap | recall (%): 39.0 -> 98.1 | Improved |
| add DTP proposals to RPN proposals (ablation) | Ctrl-F-Net · GW 15-5, 50% overlap | recall (%): 91.1 -> 99.4 | Improved |
| add RPN to DTP proposals (ablation) | Ctrl-F-Net · IAM, 50% overlap | recall (%): 97.9 -> 98.1 | No clear effect |
| use segmentation-free Ctrl-F-Net instead of segmentation-based DCToW on GW QbS | GW 15-5 25% overlap · GW QbS | MAP (%): 93.7 -> 95.2 | Improved |
| use segmentation-free Ctrl-F-Net instead of segmentation-based Deep Embedding on IAM | IAM 25% overlap · IAM QbE | MAP (%): 84.2 -> 74.1 | Hurt |
| add bigrams to PHOC alphabet (ablation) | PHOC embedding · word spotting | MAP | No clear effect |
| use Ctrl-F-Net DCToW instead of BoF HMMs | GW 5-15 50% QbS · GW 5-15 | MAP (%): 54.6 -> 73.8 | Improved |
Closest papers by meaning
- An Alternative Deep Feature Approach to Line Level Keyword Spotting (CVPR 2019 · similarity 0.87)
- Efficient Learning-Free Keyword Spotting (TPAMI 2018 · similarity 0.86)
- DAN: A Segmentation-Free Document Attention Network for Handwritten Document Recognition (TPAMI 2023 · similarity 0.86)
- Start, Follow, Read: End-to-End Full-Page Handwriting Recognition (ECCV 2018 · similarity 0.86)
- Handwritten Text Recognition: A Survey (TPAMI 2025 · similarity 0.85)
- MetaWriter: Personalized Handwritten Text Recognition Using Meta-Learned Prompt Tuning (CVPR 2025 · similarity 0.83)
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.
