Research map / Large language models / Language, safety and interpretability
MSDS: A Large-Scale Chinese Signature and Token Digit String Dataset for Handwriting Verification
Peirong Zhang, Jiajia Jiang, Yuliang Liu, Lianwen Jin
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 Token Digit Strings instead of Chinese signatures | Handwriting verification systems · MSDS-TDS vs MSDS-ChS, same-user collection conditions | EER | Improved |
| Test across sessions rather than within a single session | Handwriting verification systems · MSDS-ChS and MSDS-TDS | EER | Hurt |
| Add offline images to online time-series features (ablation) | Sig2Vec verifier · MSDS-ChS, skilled forgery 4vs1 | EER (global/local, %): 9.03/4.97 -> 8.44/4.07 | Improved |
| Add offline images to online time-series features (ablation) | Sig2Vec verifier · MSDS-ChS, random forgery 1vs1 | EER (global/local, %): 5.09/1.18 -> 3.54/1.53 | Mixed |
| Add offline images to online time-series features (ablation) | Sig2Vec verifier · MSDS-TDS, skilled forgery 4vs1 | EER (global/local, %): 5.18/2.07 -> 4.95/1.88 | Improved |
| Add offline images to online time-series features (ablation) | Sig2Vec verifier · MSDS-TDS, random forgery 1vs1 | EER (global/local, %): 1.76/0.28 -> 1.62/0.53 | Mixed |
| Use DsDTW rather than the other evaluated systems | Online handwriting verification · Skilled-forgery scenarios on MSDS-ChS and MSDS-TDS | EER | Improved |
| Use DTW rather than the other evaluated systems | Online handwriting verification · Random-forgery scenarios on MSDS-ChS and MSDS-TDS | EER | Improved |
| Train DsDTW on DeepSignDB rather than MSDS-ChS | DsDTW · MSDS-ChS, random forgery 4vs1 | EER (global/local, %): 0.84/0.11 -> 0.41/0.06 | Improved |
| Train on DeepSignDB and test on Chinese signatures | DsDTW · MSDS-ChS, skilled forgery 4vs1 | EER (global/local, %): 5.91/2.90 -> 10.60/5.78 | Hurt |
| Use four or three genuine templates rather than one | Handwriting verification systems · MSDS-ChS and MSDS-TDS | EER | Improved |
Closest papers by meaning
- Privacy-Preserving Biometric Verification With Handwritten Random Digit String (TPAMI 2025 · similarity 0.83)
- Multi-Language Online Handwriting Recognition (TPAMI 2016 · similarity 0.82)
- FSboard: Over 3 Million Characters of ASL Fingerspelling Collected via Smartphones (CVPR 2025 · similarity 0.79)
- KD-MSLRT: Lightweight Sign Language Recognition Model Based on Mediapipe and 3D to 1D Knowledge Distillation (AAAI 2025 · similarity 0.78)
- MM-WLAuslan: Multi-View Multi-Modal Word-Level Australian Sign Language Recognition Dataset (NeurIPS 2024 · similarity 0.78)
- Muharaf: Manuscripts of Handwritten Arabic Dataset for Cursive Text Recognition (NeurIPS 2024 · similarity 0.77)
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.
