DSBI
收藏资源简介:
DSBI数据集由中国科学院计算技术研究所创建,包含114张双面盲文图像,来自6本不同的盲文书和一些普通印刷文档,旨在提供多样性和复杂性。数据集通过普通平板扫描仪获取,使用200dpi分辨率捕捉彩色盲文图像,并存储为JPEG格式。创建过程中,采用了辅助标注策略,通过自动检测盲文点和便捷的人机交互方法修改标注结果,显著提高了标注效率。DSBI数据集主要应用于盲文点检测,为盲文图像识别提供核心和基础步骤,旨在推动相关研究和评估算法性能。
The DSBI dataset was developed by the Institute of Computing Technology, Chinese Academy of Sciences. It contains 114 double-sided braille images collected from 6 different braille books and some common printed documents, designed to offer sufficient diversity and complexity. The dataset was acquired using a conventional flatbed scanner, with color braille images captured at 200 dpi resolution and stored in JPEG format. During the dataset construction, an auxiliary annotation strategy was employed, which automatically detects braille dots and revises annotation results through a user-friendly human-computer interaction approach, greatly enhancing annotation efficiency. The DSBI dataset is mainly utilized for braille dot detection, a core and fundamental step in braille image recognition, with the goal of facilitating relevant research and algorithm performance evaluation.

- 1DSBI: Double-Sided Braille Image Dataset and Algorithm Evaluation for Braille Dots Detection中国科学院计算技术研究所 · 2019年



