Institutional Repository of Vilnius University
Objektų konfigūracijos atpažinimas naudojant mašininį mokymąsi /
Abstract
dc:descriptionIn order to determine suitable methods for recognizing the configuration of objects in an image, a specific problem was chosen: recognizing chess diagrams. The approach employed object detection techniques to identify both the chessboard and the individual pieces. Two models were trained using the YOLO algorithm: one for detecting chessboards and another for recognizing chess pieces. The position of the chess game is described using the Forsyth-Edwards notation (FEN). The models were improved by analyzing images that tended to be recognized incorrectly. Additionally, a graphical user interface application was developed to facilitate the recognition of chess diagrams and their transfer to chess position analysis programs. To assess accuracy, tests were conducted using image samples with varying quality, and the results were compared to other programs offering similar functionality. The application created in this work demonstrated superior performance, achieving 98 % accuracy in piece recognition and 82 % accuracy in recognizing chess game positions. These recognition results can be applied to streamline the digitization of chess publications, automate the transfer of game positions to chess-playing programs or analysis tools, and build databases of chess problems.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Paliulionytė, Alina,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192845206&prefLang=en_US