多模态签名与数字字符串数据集（MSDS）
收藏资源简介：
多模态签名与数字字符串数据集（MSDS）是由华南理工大学深度学习与视觉计算实验室创建的大型在线和离线手写数据集。该数据集包含两个子集：MSDS-ChS（中文签名）和MSDS-TDS（令牌数字字符串），由402名用户贡献，每个用户在每个子集中提供20个真实样本和20个熟练伪造样本。MSDS-ChS是目前公开可用的最大的中文签名数据集，至少是现有在线数据集的八倍大。MSDS-TDS首次涵盖了手写的令牌数字字符串，即用户的实际电话号码，这些尚未被研究过。数据集的创建过程考虑了同一用户在不同会话中的书写变化，模拟了更真实的评估场景，增强了数据集的可行性。该数据集主要用于手写身份验证的研究，旨在探索新的更有效的生物识别方法，为相关研究领域带来长期影响。
The Multimodal Signature and Digital String Dataset (MSDS) is a large-scale online and offline handwriting dataset developed by the Deep Learning and Visual Computing Laboratory of South China University of Technology. This dataset comprises two subsets: MSDS-ChS (Chinese Signatures) and MSDS-TDS (Token Digital Strings), collected from 402 distinct users, with each user providing 20 genuine samples and 20 skilled forgeries for each subset. MSDS-ChS stands as the largest publicly available Chinese signature dataset to date, with a scale at least eight times that of existing comparable online datasets. MSDS-TDS is the first dataset to cover handwritten token digital strings, specifically users' actual phone numbers, a topic that has not been explored in existing research. The construction of this dataset accounts for writing variations of the same user across different recording sessions, simulating more realistic evaluation scenarios and thereby enhancing the dataset's practical utility. This dataset is primarily utilized for research on handwriting-based identity verification, aiming to explore novel and more effective biometric approaches, and is expected to bring long-term impacts to relevant research communities.
