posted by organizer: imriss || 1025 views || tracked by 3 users: [display]

MS-TEx 2015 : ICDAR 2015 MultiSpectral Text Extraction Contest


When Aug 23, 2015 - Aug 26, 2015
Where Everywhere
Abstract Registration Due Mar 20, 2015
Submission Deadline Apr 1, 2015
Categories    multispectral   segmentation   contest

Call For Papers

In recent years, Multispectral (MS) imaging becomes a very important tool for historical document analysis. This technique is widely known as a non-invasive method of investigation thanks to its simultaneous use of ultraviolet, infrared and visible light. It enables conservators and art historians to obtain valuable information on ancient documents without causing any physical damages to the materials, and makes it possible to reveal the newly overwritten text, to distinguish and recognize the chemical material composing the ink, to detect signs of degradation in historical documents. It can also help to extract information from cultural heritage patterns which cannot be extracted using conventional color photography. Extracting (segmenting) the original text (old writing) from MS document image is a very important step for subsequent document image analysis and investigation. In order to facilitate comparison of the results of different algorithms and track their progress over time towards a level of human performance, it would be of great interest to obtain a standard benchmark and accurate ground-truth with the most representative information about the targeted samples. To this end, we have generated a dataset of 21 MS document images for training purposes (, and another dataset of ten MS document images for testing purposes.

Related Resources

SLIE 2017   Semantic, Logics, Information Extraction and AI
TMNZ 2016   First New Zealand Text Mining Workshop
SLIE 2016   Semantic, Logics, Information Extraction and AI (SLIE)
SENTIRE 2016   6th ICDM Workshop on Sentiment Elicitation from Natural Text for Information Retrieval and Extraction
PolText 2016   International Conference on the Advances in Computational Analysis of Political Text
LD4IE 2016   CFP: Linked Data for Information Extraction LD4IE2016 - workshop at @ISWC2016
EGC 2017   ConfĂ©rence Extraction et Gestion des Connaissances (EGC 2017)
WNUT 2016   The 2nd Workshop on Noisy User-generated Text @COLING
ADCS 2016   The 21st Australasian Document Computing Symposium
MAMLAKE 2017   Special Session on Modern Applications of Machine Learning for Actionable Knowledge Extraction