Real-Time Chessboard State Recognition Using YOLOv8: A Comprehensive Approach to Chess Pieces Detection and Analysis
Keywords:
Augmented Reality, Automated Chess Analysis, Chessboard Recognition, Computer Vision, Kaggle Dataset, Machine Learning, Real-Time Object Detection, YOLOv8Abstract
This research presents a cutting-edge approach to real-time chessboard state recognition using the YOLOv8 object detection framework. Chess, known for its strategic depth, requires precise analysis of board states for gameplay enhancement and AI-driven insights. Leveraging a carefully curated Kaggle dataset comprising diverse chess board configurations and lighting conditions, this study develops a robust system capable of accurately detecting and classifying chess pieces in real-time scenarios. The model's architecture integrates an anchor-free mechanism, multi-scale detection capabilities, and advanced training strategies like mosaic augmentation and adaptive learning rates, ensuring optimal performance.
The detection system achieves an impressive mAP@50 IoU of 98.7%, with frame rates surpassing 30 FPS on Tesla T4 GPUs, making it ideal for seamless real-time applications. Evaluation metrics, including precision, recall, and qualitative analyses, confirm the system's reliability under diverse conditions, such as occlusions and varying perspectives. Potential applications span automated gameplay analysis, augmented reality learning tools, and portable edge-device deployments.
The study also outlines future directions, including expanding the dataset, integrating detection with chess engines, and optimizing for mobile platforms. By bridging traditional chess gameplay and AI, this research highlights the transformative potential of YOLOv8 in enhancing the chess-playing experience and contributing to technological advancements in the domain.
References
P. Ranasinghe, P. Ranasinghe, and V. Ashan, “ChessEye: An Integrated Framework for Accurate and Efficient Chessboard Reconstruction,” 2022 Moratuwa Engineering Research Conference (MERCon), pp. 177–182, Nov. 2023, doi: https://doi.org/10.1109/mercon60487.2023.10355515.
P. Yarmohammadi, N. A. Khomami, M. T. Masouleh, and M. R. Zakerzadeh, “Experimental Study on Chess Board Setup Using Delta Parallel Robot Based on Deep Learning,” 2023 11th RSI International Conference on Robotics and Mechatronics (ICRoM), pp. 869–875, Dec. 2023, doi: https://doi.org/10.1109/icrom60803.2023.10412348.
X. Liu, J. Jin, Q. Zhong, Q. Lei, Z. Wu, and Z. Fang, “Vision-Based Chess Detection for a Robotic Companion,” 2019 IEEE 4th International Conference on Computer and Communication Systems (ICCCS), vol. 37, pp. 248–253, Apr. 2021, doi: https://doi.org/10.1109/icccs52626.2021.9449305.
Y. Jin, “Design and Implementation of Chess and Card Recognition System Based on Deep Learning,” Frontiers in Computing and Intelligent Systems, vol. 2, no. 1, pp. 96–100, Nov. 2022, doi: https://doi.org/10.54097/fcis.v2i1.3171.
M. Wang, “Vision-based Robotic Chess Player,” Washington.edu, Aug. 26, 2021. https://digital.lib.washington.edu/researchworks/items/d8db1e76-8a46-446d-9f73-33787e6139c6 (accessed Jan. 16, 2025).
A. I. Bugarin, “Real Time Tracking and Analysis of Physical Chess Games Using Computer Vision and Machine Learning,” Escholarship.org, Jul. 24, 2024. https://escholarship.org/uc/item/31d7k315 (accessed Jan. 16, 2025).
S. P. Yadav, M. Jindal, P. Rani, V. Hugo, S. Nascimento, and M. Kumar, “An Improved Deep Learning-Based Optimal Object Detection System from Images,” Multimedia Tools and Applications, Sep. 2023, doi: https://doi.org/10.1007/s11042-023-16736-5.
M. Conner and R. Talvi, “YOLO for Chess Detection An Implementation of the YOLO Algorithm,” 2023. Available: https://dothereading.github.io/projects/YOLO_Chess.pdf
“Proceedings of Ninth International Congress on Information and Communication Technology: Icict 2024, London, Volume 10 (Lecture Notes in Networks and Systems #1055),” IndieCommerce, 2024. https://icdemo.indiecommerce.com/book/9789819754403?v=43802 (accessed Jan. 16, 2025).
J. Schrittwieser et al., “Mastering Atari, Go, chess and shogi by planning with a learned model,” Nature, vol. 588, no. 7839, pp. 604–609, Dec. 2020, doi: https://doi.org/10.1038/s41586-020-03051-4.
