Tesseract OCR of IIT-CDIP Dataset

This is Tesseract generated <strong>transcriptions (no images)</strong> of (most of) the IIT-CDIP dataset. To download the images of the IIT-CDIP dataset go to https://data.nist.gov/od/id/mds2-2531 The directory struture of this dataset is the same as the IIT-CDIP dataset (although has everything in one tar, with "a.a", "a.b", ... directories) and can thus be combine with the image IIT-CDIP dataset using rsync or similar tool. This dataset contains a "X.layout.json" for each "X.png" in the IIT-CDIP dataset (doesn't have sections 'a', 'w', 'x', 'y', and 'z'). The jsons contain block/paragraph, line and word bounding boxes, with transcriptions for the words following the Tesseract format. The line and word annotations are directly taken from Tesseract. The block and paragraph output of Tesseract was discarded. The images were then run through both the Publaynet and PrimaNet models available on LayoutParser (https://layout-parser.github.io/). The combine output of these models became the block/paragraph annotations (we kept the Tesseract output format, but each block has 1 paragraph of exactly the same shape). <strong>Important:</strong> There is also a "rotation" value in the json (0, 90, 180, or 270) indicating the json may be for a rotated version of the IIT-CDIP image by the given amount (attempted to rotated documents to upright position to get better OCR results). These are the annotations used to pre-train <strong>Dessurt</strong> (https://arxiv.org/abs/2203.16618). These annotations will be worse than those that would be obtained using a commercial OCR system (like those used to pre-train LayoutLMv2/v3). The code used to produce these annotations is available here: https://github.com/herobd/ocr本数据集为**Tesseract生成的转录文本（无图像）**，覆盖绝大多数IIT-CDIP数据集。若需下载IIT-CDIP数据集的原始图像，请访问：https://data.nist.gov/od/id/mds2-2531。本数据集的目录结构与IIT-CDIP数据集一致（所有内容打包于单个tar文件中，包含"a.a"、"a.b"等子目录），因此可通过rsync或同类工具与IIT-CDIP图像数据集进行合并。本数据集为IIT-CDIP数据集中每个"X.png"文件对应一个"X.layout.json"文件（不包含'a'、'w'、'x'、'y'、'z'分区）。JSON文件包含文本块/段落、行级与词级边界框，以及遵循Tesseract格式的单词转录结果。其中行与词级标注直接取自Tesseract输出，而Tesseract的文本块与段落输出已被舍弃。随后我们使用**LayoutParser**（https://layout-parser.github.io/）提供的Publaynet与PrimaNet模型对原始图像进行推理，将两个模型的融合输出作为文本块/段落标注（我们保留了Tesseract的输出格式，但每个文本块仅包含一个形状完全一致的段落）。**重要提示：** JSON文件中还包含一个"rotation"字段，取值为0、90、180或270，用于指示该JSON对应的IIT-CDIP图像已被旋转对应角度——我们尝试将文档旋转至正向以获得更优的**光学字符识别（OCR）**结果。本数据集的标注被用于预训练**Dessurt**（https://arxiv.org/abs/2203.16618）。相较于商用光学字符识别（OCR）系统（如用于预训练LayoutLMv2/v3的OCR系统）生成的标注，本数据集的标注质量相对逊色。用于生成本数据集标注的代码已公开，获取地址为：https://github.com/herobd/ocr