Generating FEN Descriptions of Chess Boards by Using Two Important Pre-trained CNN Architectures
This project was made as the term project of the BLG 506E Computer Vision master course of ITU in 2021 spring term.
All presentations, project proposal and project report can be found in this repo.
If you want to learn more information about the project, you can read project proposal documents or the final report.
The dataset used in this project belongs to the Chess Positions dataset put by Pavel Koryakin on the Kaggle website.
Please visit the https://www.kaggle.com/koryakinp/chess-positions to download the dataset.
The whole dataset includes 100000 images of a randomly generated chess positions of 5-15 pieces. For this project, it is planned to use less amount of images that will be randomly selected for train, validation and test datasets.
If you want to use any information or code snippet from the project, please reference me and this project. Also, please give a star to this repo.