Přehled o publikaci

The subject of interest is automatic optical character recognition (OCR). While the OCR of ordinary printed documents has already been sufficiently mastered using high-quality OCR tools, tens of researchers and experimenters only in recent years. So far, for the Czech manuscript (kurent), there is only one historical script transcription model in the Transkribus platform, but it is based on very small script patterns. This means that a sufficiently effective aggregated model is not yet available in the Czech Republic automatic transcription that would be built on a large enough number of Ground Truth quality pages that could be used to build better transcription models. The consequence of the absence of automatic transcription tools is that historical documents of libraries, museums, archives, etc. they are though digitized, but are usually only available as digital facsimiles, images (digitized) without transcription. This scientific task is more a task for national institutions... than for SGS. The aim of this small SGS project is to contribute to solving the problem of transcription and to prepare experts who will gradually become solve this important task regarding making available historical documents from Czech archives, libraries, museums, etc. Taking into account the language skills of the researcher Vojtěch Říha, the most relevant option is to use the Czech language, so we would focus on the creation of the HTR model in the Czech language. Another much more important reason is the current absence of such a model on the Transkribus platform, which we would use for research, which is also why it would be beneficial to have a high-quality HTR model available for further potential automatic transcription of handwritten documents written in the Czech language. In the Czech language, there is still no comprehensive guide for the creation of the HTR model on the Transkribus platform, so this question of the correct procedure would also be the content of the work. Working draft of the title: • Aspects and principles applied in the creation of the HTR model intended for the automatic transcription of handwritten documents written in the Czech language (Sedlnická chronicle or other Czech manuscript documents). Lukáš Němec will devote himself to the creation of a transcription model based on historical documents and the researcher's cooperation with the Silesian Museum in Opava. It will focus on the topic of digitization, digitization and transcription of documents from the Capucin monastery library with the Transkribus tool, to transcribe the text and describe - the methodology of the work.