Face Recognition System of Professional Esports Players Based on Convolutional Neural Network

Face recognition is a complex topic in computer vision and deep learning. It is becoming even more important in various fields, such as professional esports. This paper proposes a system designed to accurately identify professional esports players during real-time translations of tournaments, competitions, and interviews. On the way to the development of this system can meet various challenges. Accordingly, this paper examines computer vision algorithms as a reliable way to solve them. From a deep learning perspective, the Convolutional Neural Network (CNN) will be the primary algorithm and the recognition accuracy of the system will depend on the training accuracy, training loss, validation accuracy, and validation loss. This paper involved 5 professional esports players, each with 99 to 112 images used for training and the same number of images used for validation. The result obtained on 100 epochs gives a best accuracy of 85%.