HASYv2
收藏资源简介:
HASYv2数据集是由Karlsruhe Institute of Technology创建的公开免费手写符号数据集,类似于MNIST。该数据集包含168,233个实例,涵盖369个类别,包括阿拉伯数字和拉丁字符等。数据集的创建过程涉及从HWRT数据集转换,并通过手动检查确保质量。HASYv2适用于训练非草书手写文档的语义分割模型,如数学笔记或表格,旨在解决手写符号识别的挑战。
The HASYv2 dataset is a publicly accessible and free handwritten symbol dataset developed by Karlsruhe Institute of Technology, which is analogous to the MNIST dataset. It contains 168,233 instances across 369 categories, including Arabic numerals, Latin characters and other related symbol types. The dataset was constructed by converting data from the HWRT dataset, with its quality ensured through manual inspection. The HASYv2 dataset is suitable for training semantic segmentation models for non-cursive handwritten documents such as mathematical notes or tables, and it aims to address the challenges associated with handwritten symbol recognition.




