READ revolutionizes access to handwritten documents

From the Middle Ages to today, from old Greek to modern English, from running text to tables or forms

About

READ's mission is to revolutionize access to archival documents with the support of cutting-edge technology such as Handwritten Text Recognition (HTR) and Keyword Spotting (KWS).

Learn more

Network

READ addresses archives and libraries, humanities scholars, family historians, volunteers - and computer scientists

Learn more

Research

Research in READ comprises exciting fields such as Artificial Intelligence, Pattern Recognition, Machine Learning and Natural Language Processing.

Learn more

Services

READ technology is available via the service platform Transkribus. Upload documents, train a Handwritten Text Recognition (HTR) model, process text and follow the progress of the project.

Learn more

Recent Posts

+ Bringing archives and technology together!

The READ project is featured in the latest issue of Insights magazine, which is produced by the ICARUS organisation.  ICARUS represents more than 160 archives in over 30 countries.

The coverage explains the aims of the project and also gives a summary of the ‘Technology meets Scholarship’ conference in January 2016, where the READ project held its kick-off meeting.

Our appearance in Insights magazine should help to spread the word about READ!  We look forward to developing our relationship with the archival community as our project continues.

+ National Archives Finland takes first steps towards Handwritten Text Recognition

The National Archives of Finland is committed to promoting access to documents relating to Finland’s cultural heritage.  Handwritten Text Recognition (HTR) technology is now part of its mission.

The National Archives of Finland has digitised millions of documents, most of which are handwritten.  As a first step, 500 of these digitised pages have now been uploaded and transcribed in the Transkribus platform.  These documents range from the sixteenth to the nineteenth century and include estate inventories of the Finnish nobility, court books and land tax registers.  These 500 pages represent training data and will play a vital role in enabling HTR engines to recognise Swedish handwriting (which was used in official documents in Finland at this time).

Aatelin_perukirjat_65

Manuscript page from the records of the Turku Court of Appeal, 1828-1829  (Image courtesy of Digital Archive, National Archives Finland)

READ researchers will use these pages to generate a HTR model that can be applied to other documents held in the National Archives of Finland.  This model will make it possible to automatically transcribe and search images of historical manuscripts, thereby ensuring easier access to the records of Finnish history.

500 pages is just the beginning!  The National Archives of Finland will continue to contribute training data as we move through the READ project and this data will help to improve the accuracy of the HTR technology.