Institutional Repository of Vilnius University
Šachmatų lentos ir figūrų atpažinimas naudojant giliuosius neuroninius tinklus /
Abstract
dc:descriptionChessboard and chess pieces recognition is a problem which is yet to be solved efficiently. Solution to this problem would be beneficial for organizers of chess tournaments as well as amateur players who want to analyse their game position on the internet with the help of artificial intelligence. Usually, this task is done by people themselves, but there already are several solutions which automated it with great results. However, this task leaves little room for error, and currently none of the proposed solutions work perfectly, so in this paper a new method is proposed. Other similar works have been analysed in this paper and it was decided to change the way chess pieces are detected – instead of dividing input image into squares and recognizing each of them separately, state of the art object detection algorithms are used which can recognise pieces from the whole image. A new dataset was created and labeled to train this model. The final product is a system which assigns found pieces to their respective squares and prints the result in FEN notation. Proposed solution achieved good results and showed the potential of such system.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Juknonis, Haroldas,
- Contributors dc:contributor
-
- Valaitis, Vytautas
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:ELABAETD107116168&prefLang=en_US