The ADOCHS project is a four-year project dedicated to the improvement of the quality control process concerning the digitised heritage collections. The goal is to address the issue in two stages. First, by focusing on methodological aspects, proposing guidelines applicable in the treatment of heritage collections. Secondly, by developing technical tools that automate tasks related to quality control or support the manual check. In addressing this issue, the project aims to ensure to speed up the whole digitising process while increasing the integrity and consistency of files and data produced.
In Belgium, the first digitisation plan has been launched in the 2004. The issue of quality control was one of the major obstacles encountered. In order to define a clear procedure of quality control to be followed in the heritage institutions, it is expected to work on two use cases: the iconographic documents of the CegeSoma and the textual documents of KBR.
Focus 1: Image Quality
The first step will include conducting a literature study on image quality assessment and identifying problems encountered by the institutions. Based on our findings, new objective perceptual quality metrics and quality enhancement methodologies will be developed tuned towards our two use cases.
Focus 2: Metadata Quality
Simultaneously to a literature study on metadata quality, a typology of frequently occurring problems with manually encoded and automatically generated metadata will be developed. These preliminary results will allow us to establish new metrics as well as a workflow reusable by other institutions.
Researchers Tan Lu (VUB/KBR) and Ann Dooms (Cegesoma) won the ICDAR 2019 competition in document recognition with complex layouts. Read moreMore information about the project