AIKON platform

Aikon is a modular computer vision platform that enables historians to build, process, and analyze visual corpora at scale. The platform guides users through a complete workflow from corpus construction to algorithmic processing and result validation, without requiring technical expertise. Built on IIIF standards and featuring a flexible data model, Aikon supports collaborative research while maintaining full user control over automatic processing. Its modular architecture allows easy integration of new computer vision algorithms and visualization tools, making it adaptable to diverse research needs across historical document analysis.

This repository contains the code for the frontend platform, as well as a submodule for the worker API.

General requirements

• Sudo privileges
• Python == 3.10
• Git:
  • sudo apt install git
  • Having configured SSH access to GitHub

Install 🛠️

Please refer to front/README and api/README for detailed instructions (especially step-by-step install).

1. To install the API and Front application inside the front and api folders, run:

   bash setup.sh

2. Define the .env variables to fit your requirements: For front application (front/app/conf/.env), notably

   # Folder where the media files are stored
   MEDIA_DIR=/home/path/to/aikon/front/app/mediafiles

   For the API (api/.env), notably

   # Folder where the data is stored
   API_DATA_FOLDER=data/

3. To start everything in one killable process, run (after installing each part like advised in the subfolders):

   bash run.sh

Acknowledgements

Aikon is funded and supported by the Agence Nationale pour la Recherche and the European Research Council

• VHS ANR-21-CE38-0008: computer Vision and Historical analysis of Scientific illustration circulation
• EiDA ANR-22-CE38-0014: EdIter et analyser les Diagrammes astronomiques historiques avec l’intelligence Artificielle
• DISCOVER project ERC-101076028: Discovering and Analyzing Visual Structures

If you find this work useful, please consider citing:

@article{albouy2025aikon,
    title={{AIKON: A Modular Computer Vision Platform for Historical Corpora}},
    author={
        Albouy, Ségolène and
        Norindr, Somkeo and
        Kervegan, Paul and
        Aouinti, Fouad and
        Delanaux, Rémy and
        Champenois, Robin and
        Grometto, Clara and
        Lazaris, Stavros and
        Guilbaud, Alexandre and
        Husson, Matthieu and
        Aubry, Mathieu
    },
    url={https://hal.science/hal-05248250},
    year={2025},
    month={Sep},
    number={hal-05248250},
    journal={HAL Pre-Print},
    keyword={Digital Humanities, Computer Vision, Historical Documents, Visual Analysis},
}