huridocs/pdf-document-layout-analysis

A Docker-powered service for PDF document layout analysis. This service provides a powerful and flexible PDF analysis service. The service allows for the segmentation and classification of different parts of PDF pages, identifying the elements such as texts, titles, pictures, tables and so on.

What it solves

This project provides a microservice for intelligent PDF layout analysis, allowing users to convert complex PDF documents into structured formats like Markdown or HTML while preserving the reading order and identifying specific content types (such as titles, tables, and formulas).

How it works

The service uses a combination of machine learning models to segment and classify PDF content. It offers a choice between the Vision Grid Transformer (VGT) for high accuracy and LightGBM for faster processing. It integrates Tesseract OCR for support of over 150 languages and uses Ollama-powered models to provide automatic translation of the extracted content into multiple target languages.

Who it’s for

It is designed for developers and organizations needing to programmatically extract structured data from PDFs, as well as users who prefer a visual interface via a Gradio web UI for document conversion and translation.

Highlights

  • Multi-model Analysis: Choose between high-accuracy VGT or high-speed LightGBM models.
  • Structured Export: Convert PDFs to Markdown or HTML, including LaTeX for formulas and HTML for tables.
  • Integrated Translation: Automatic document translation using Ollama models while preserving formatting.
  • Comprehensive Tooling: Includes a REST API, a Gradio web interface, and Docker support with GPU acceleration.
  • OCR Capabilities: Supports 150+ languages with features for page rotation and deskewing.

Related

  • Project
    open-edge-platform/geti

    An end-to-end platform for building computer vision AI models, providing tools for annotation, training, and optimization for deployment on edge hardware.

  • Project
    ndl-lab/ndlocr-lite

    A lightweight OCR tool designed for home computers to extract text from digitized books and magazines without requiring a GPU.

  • Project
    facebook/ThreatExchange

    A collection of tools and APIs for content moderation and security threat intelligence, featuring image and video hashing algorithms for similarity matching.

  • Project
    zju3dv/street_gaussians

    Street Gaussians is a research‑grade codebase that extends 3‑D Gaussian splatting to reconstruct and render dynamic street‑level scenes (e.g., Waymo data). It provides data‑conversion tools, training/evaluation scripts, and optional LiDAR‑depth/sky‑mask preprocessing, all wrapped in configurable YAML pipelines.