1,000张日语发票采集数据

1,000张日语发票采集数据。数据包含基础编辑500张，专业编辑500张。数据多样性包括不同发票内容、不同编辑类型、多种发票格式。发票上的公司名字、地址、姓名、传真号、电话等敏感信息均进行了虚拟化编辑，非真实信息。数据可用于日语发票检测、识别、端到端OCR等任务。

This dataset consists of 1,000 collected Japanese invoice samples, including 500 basic-edited samples and 500 professionally-edited samples. The data features diversity in terms of invoice contents, editing types and invoice formats. All sensitive information on the invoices, such as company names, addresses, contact persons, fax numbers and phone numbers, has been virtualized and does not correspond to real entities. This dataset can be used for tasks including Japanese invoice detection, recognition and end-to-end OCR.

搜集汇总

背景与挑战

背景概述

该数据集包含1000张日语发票图像，由500张基础编辑和500张专业编辑构成，涵盖了不同的发票内容、编辑类型和格式，所有敏感信息均已进行虚拟化处理。它适用于日语发票检测、识别以及端到端OCR等人工智能任务。