Paper Accepted @ IEEE Transactions on Games
IEEE Transaction on Games accepted my paper From Pixels to Titles: Video Game Identification by Screenshots using Convolutional Neural Networks.
The paper evaluates the ability to identify video games using screenshots with different CNN and transformer architectures. EfficientNetV2S showed the highest average accuracy at 77.44%, outperforming other models in video game identification.
The preprint is available here.
Some examples of screenshots from the proposed dataset.
Fabricio Breve
Fabricio Breve received his bachelor's degree from the Methodist University of Piracicaba, Brazil in 2001, his master's degree from the Federal University of Sao Carlos, Brazil in 2006, and his Ph.D. from the University of Sao Paulo, Brazil in 2010, with a collaborative period at the University of Alberta, Canada. He is currently an associate professor at Sao Paulo State University, Brazil. His research interests include machine learning, pattern recognition, image processing, artificial neural networks, complex networks, and nature-inspired computing.
Permanent link to this article: https://www.fabriciobreve.com/2025/01/08/paper-accepted-ieee-transactions-on-games/
