Training set directory

This file contains the training set. It has 6 subdirectories with 5.000 invoices in each one. There are three files per invoice: invoice000(n).pdf has the content of bill (n) in PDF format; invoice000(n).json is the file with the true labels; and invoice000(n)_ann.pdf is the bill with annotations (the name of labels are placed around the corresponding value in the PDF file).

本文件包含训练集。该数据集共包含6个子目录，每个子目录内含5000份发票单据。每份发票对应三个文件：invoice000(n).pdf 为以PDF格式存储的第n份发票单据内容；invoice000(n).json 为存储真实标签的文件；invoice000(n)_ann.pdf 为带有标注的发票单据（标签名称标注于PDF文件中对应数值的周边位置）。