HIMANIS Project

Goals

Manuscripts are among the most important witnesses to our European shared cultural heritage. Despite a large digitization, the wealth of their content remains largely inaccessible : current handwritten text recognition technology is not accurate enough to allow full text search. This situation raises the need of searching approaches specifically designed for text images.

HIMANIS aims at developing cost-effective solutions for querying large sets of handwritten document images.

Transcription and Indexing

a new indexing/searching technology for historical manuscripts

Full text search

a new paradigm to study our historical heritage, as conveyed by manuscripts, by using full text search technology.

History of Europe

a new vision of the raise of nation states in Europe via a new study of the corpus under this paradigm.

The Chancery Corpus

The “Tresor des Chartes” is the most valuable and iconic archive of the French kings and has been built as a collection as early as in the 13th c. Within the “Tresor”, the collection of medieval registers (aka “Chancery corpus”) stand out. They keep record of the charters, grants and privileges given by the king of France and were produced directly by the French royal chancery. Dating from 1302 to 1483, they contain ca. 68,000 charters and documents. This large and iconic collection bears witness to the rationalization of late medieval administration and is a key source to our understanding of medieval Europe and the rise of centralized nation state on the continent as a consequence of the long lasting wars between France and England.

HIMANIS Search engine

The 60,000 most relevant pages of the Trésor des Chartes have been made searchable thanks to a novel indexing technique specially designed in HIMANIS for manuscript images. A simple, preliminary interface to this index is provided here.

Enter your query

you can search for individual words and word combinations. More details about how to use this search interface, along with examples of queries, can be found here

Members

Institut de recherche et d’histoire des textes

Paris, France

CNRS institute devoted to fundamental research on medieval manuscripts and early printed books.

A2iA

Paris, France

The global leader in handwriting recognition, text extraction and document classification software.

Pattern Recognition and Human Language Technology

Valencia, Spain

Universitat Politècnica de València research center dedicated to Multimodal Interaction, Pattern Recognition, Image Processing and Language Processing

Artificial Intelligence and Cognitive Engineering

Groningen, The Netherlands

Research center covering Cognitive Modeling, Multi-Agent Systems, Autonomous Perceptive Systems, Sensory Cognition.

Partners

Teklia

France

Machine Learning and Deep Learning Agency specialized on Data Science and Data Viz

Archives Nationales

France

The French national Archive public service

Bibliothèque Nationale de France

France

Collect, preserve, enrich and make available the French national documentary heritage

Publications

On HIMANIS corpus

Preparatory KWS Experiments for Large-Scale Indexing of a Vast Medieval Manuscript Collection in the HIMANIS Project., DOI: 10.1109/ICDAR.2017.59, 2017

On other corpora

HMM word graph based keyword spotting in handwritten document images., DOI: 10.1016/j.ins.2016.07.063, 2016
Word graphs size impact on the performance of handwriting document applications., DOI: 10.1007/s00521-016-2336-2, 2017
Handwriting Transcription and Keyword Spotting in Historical Daily Records Documents., DOI: 10.1109/DAS.2016.70, 2016
Two Methods to Improve Confidence Scores for Lexicon-free Word Spotting in Handwritten Text., DOI: 10.1109/ICFHR.2016.0072, 2016
Querying out-of-vocabulary words in lexicon-based keyword spotting., DOI: 10.1007/s00521-016-2197-8, 2017
Simple and Effective Multi-word Query Spotting in Handwritten Text Images., DOI: 10.1007/978-3-319-58838-4_9, 2017
Overview of the ImageCLEF 2016 handwritten scanned document retrieval task., DOI: 10.5281/zenodo.52994, 2016
ICFHR2016 Handwritten Keyword Spotting Competition (H-KWS 2016), DOI: 10.1109/ICFHR.2016.0117, 2016