Image Collation

(ICDAR 2021) Pytorch implementation of Paper "Image Collation: Matching illustrations in manuscripts"

[PDF [Project website] [1min Video] [Slides]

The project is an extension work to ArtMiner and Historical Watermark Recognition. If our project is helpful for your research, please consider citing :

@inproceedings{kaoua2021imagecollation,
  title={Image Collation: Matching illustrations in manuscripts},
  author={Kaoua, Ryad and Shen, Xi and Durr-Lazaris, Alexandra and Lazaris, Stavros and Picard, David and Aubry, Mathieu},
  booktitle={International Conference on Document Analysis and Recognition (ICDAR)},
  year={2021}
}

Table of Content

• Installation
• Download Manuscript
• Match Illustration
• Visualize Matches

Installation

Code is tested under Pytorch > 1.0 + Python 3.6 environment. To install all dependencies :

bash requirement.sh

Download Manuscript

ManuscriptDownloader contains a script that will download the manuscripts' folios for you, and extract the illustrations from them.

Match Illustration

IllustrationMatcher will run our algorithm to match two manuscripts. The results can be visualised with ManuscriptTools

Visualize Matches

Given two manuscripts, and their respective illustrations extracted, ManuscriptTools will generate a set of web pages.