arXiv 2201.066482022-01-17cs.CVcs.LG

OmniPrint：一种可配置的印刷体字符合成器

OmniPrint: A Configurable Printed Character Synthesizer

LISN (CNRS/INRIA) Université Paris-Saclay（巴黎-萨克雷大学 LISN（法国国家科学研究中心/法国国家信息与自动化研究所））；Paradigm Inc（第四范式公司）；ChaLearn

Haozhe Sun, Wei-Wei Tu, Isabelle Guyon

备注：: Accepted at 35th Conference on Neural Information Processing Systems (NeurIPS 2021) Track on Datasets and Benchmarks. https://openreview.net/forum?id=R07XwJPmgpl
发表信息：: 35th Conference on Neural Information Processing Systems (NeurIPS 2021) Track on Datasets and Benchmarks

摘要：

我们介绍OmniPrint，一个孤立的印刷字符的合成数据生成器，面向机器学习研究。它从MNIST、SVHN和Omniglot等著名数据集中汲取灵感，但提供了从各种语言、字体和样式生成各种印刷字符的能力，并具有自定义扭曲。我们包括来自27个脚本和许多类型的扭曲935字体。作为概念验证，我们展示了各种用例，包括为即将到来的MetaDL NeurIPS 2021竞赛设计的元学习数据集示例。OmniPrint可在https://github.com/SunHaozhe/OmniPrint上获得。

英文摘要：

We introduce OmniPrint, a synthetic data generator of isolated printed characters, geared toward machine learning research. It draws inspiration from famous datasets such as MNIST, SVHN and Omniglot, but offers the capability of generating a wide variety of printed characters from various languages, fonts and styles, with customized distortions. We include 935 fonts from 27 scripts and many types of distortions. As a proof of concept, we show various use cases, including an example of meta-learning dataset designed for the upcoming MetaDL NeurIPS 2021 competition. OmniPrint is available at https://github.com/SunHaozhe/OmniPrint.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。