In this application, a methodology is demonstrated that employs word spotting technique to speed up the manual transcription process and reduces the creation cost of training data.
Keyword spotting (KWS) can be defined as the task of identifying locations on a document image which have high probability to contain an instance of a queried word, without explicitly recognizing it.
The main advantage of KWS systems is that they only perform word detection without any training data or any language model, and thus it makes them ideal as recommender system for helping the user transcribe the document.
The major achievement of the current application is the reduction in time expenses required to achieve transcription data which could feed a Handwriting Text Recognition engine in a bootstrapping fashion. The keyword spotting pipeline is coupled with a relevance feedback mechanism which introduces the user in the retrieval loop, thus, improving the final retrieval performance.
Current work is supported by the European Unions Seventh Framework Programme (FP7/2007-2013) under grant agreement No. 600707 - tranScriptorium.
More details: http://transcriptorium.eu/
Developer: Konstantinos Zagoris (http://www.zagoris.gr)
Visual Computing Group (http://vc.ee.duth.gr/)
National Center for Scientific Research \"Demokritos\" (http://www.demokritos.gr/?lang=en)