Convert documents to structured data effortlessly. Unstructured is open-source ETL solution for transforming complex documents into clean, structured formats for language models. Visit our website to learn more about our enterprise grade Platform product for production grade workflows, partitioning, enrichments, chunking and embedding.

A Unified Toolkit for Deep Learning Based Document Image Analysis

A comprehensive list of awesome document image rectification papers.

The official code for “DocTr: Document Image Transformer for Geometric Unwarping and Illumination Correction”, ACM MM, Oral Paper, 2021.

The official repo for “DocScanner: Robust Document Image Rectification with Progressive Learning”, IJCV, 2025.

Detectron2 for Document Layout Analysis

The official code for “Geometric Representation Learning for Document Image Rectification”, ECCV, 2022.

文档图像处理工具(Document image processing tool)，包括漂白 / 文字方向矫正 / 清晰增强 / 笔记去噪美化 / 去阴影 / 扭曲矫正 / 切边增强(DocBleach / TextOrientationCorrection / DocSharpening / HandwritingDenoisingBeautifying / DocShadowRemoval / document_image_dewarping / DocTrimmingEnhancement)。

Android App for English Handwritten Text Recognition

Process Caltech Archives' digital documents and photos, and annotate each page or image with information about its contents

The ScriptNet / competitions site.

Python wrapper to facilitate data manipulation for the SmartDoc 2015 - Challenge 1 Dataset.

A web app evaluating the quality the scanned document images

复杂背景图像漂白，文字方向矫正，清晰增强，笔记去噪美化，去阴影，扭曲矫正，去黑点以及切边增强。complex background image bleaching, text direction correction, clarity enhancement, note to blur beautification, shadow removal, distortion correction, black spots removal and cutting edge enhancement。

Open source libraries and APIs to build custom preprocessing pipelines for labeling, training, or production machine learning pipelines.

This script automates the process of extracting text from various file formats (images, PDFs, DOCX) using Optical Character Recognition (OCR) powered by Azure Cognitive Services. The script supports image preprocessing, text extraction, and uploading of the processed files to Google Cloud Storage (GCP).

Sophia Trikoupi dataset (Collection of 46 handwritten, annotated pages)