Laboratoire de Recherche LATIS Laboratory of Advanced Technology and Intelligent Systems


ICDAR 2019 Competition on Historical Book Analysis
02 AVR
ICDAR 2019 Competition on Historical Book Analysis

We invite you to participate in our challenges on Historical Book Analysis (HBA) Competition in the context of the 15th IAPR International Conference on Document Analysis and Recognition (ICDAR’19).


We remind you that the registration of interest is open until April 30, 2019. 


Register your interest here (already 31 registered participants)



In conjunction with the 15th IAPR International Conference on Document Analysis and Recognition ICDAR’19, the HISTORICAL BOOK ANALYSIS COMPETITION (HBA) is organized. The HBA competition will address a thriving topic of major interest of many researchers in different fields including (historical) document image analysisimage processingpattern recognition and classification.


The HBA competition will provide a large experimental corpus and a thorough evaluation protocol to ensure a consistent comparison of image processing methods for historical document image analysis.


A challenging dataset called the HBA 1.0 dataset will be used at this occasion. The HBA 1.0 dataset is composed of 4436 real scanned ground-truthed one-page historical document images (2435 and 2001 manuscript and printed pages, respectively.) from 11 books (5 manuscripts and 6 printed books) in different languages and scripts published between the 13th and 19th centuries.

The documents of the HBA 1.0 dataset are gray-scale or color images which were digitized at 300 or 400 dpi and saved in the TIFF format which provides a high resolution of digitized images. Each selected foreground pixel is marked by a color that symbolizes the corresponding content type. The ground truth information is currently available at the pixel level.


Two nested challenges are proposed in the HBA competition.

   1- The HBA competition will aim at evaluating how image analysis methods could discriminate the textual content from the graphical ones at pixel level.

   2- It will aim at assessing the capabilities of the participating methods to separate the textual content according to different text fonts (e.g. lowercase, uppercase, italic, …) at pixel level.

How to participate

   1- Register your interest through the registration form (before 30 April 2019).

   2- Specify clearly in which challenge you would like to participate (challenge 1 only, challenge 2 only, both challenges 1 and 2).

   3- Download the sample dataset (available from 10 January 2019).

   4- Download the evaluation dataset (available from 01 March 2019).

   5- Submit the description and the results of your methods (before 31 May 2019).

Important dates



January 10, 2019

– Opening of the registration to competition

– Publication of the sample dataset

April 30, 2019

– Closure of the registration to competition

– Publication of the evaluation dataset

– Beginning of the competition

May 31, 2019

– Deadline of the result submission

– Deadline of the submission of the description of the participating methods

June 15, 2019

– Sending the competition results to the participants


– Maroua Mehri[1]

– Pierre Héroux[2]

– Rémy Mullot[3]

– Jean-Philippe Moreux[4]

– Bertrand Coüasnon[5]

– Bill Barrett[6]


[1] LATIS, Sousse University, Tunisia

[2] LITIS, University of Rouen Normandy, France

[3] L3i, University of La Rochelle, France

[4] BnF - French national library, France

[5] Intuidoc, IRISA Rennes, France

[6] Family History Technology Lab, Brigham Young University, USA 


ICDAR 2019 Competition on Historical Book Analysis


See more information at our website:

Contact us on:

Follow us on Twitter: @icdar2019hba  

Download the call for participation here: