South Korea's Gwangju Institute of Science and Technology (GIST) have developed a framework for AI-powered in-game observers for esports tournaments.
Follow Us
Researchers at South Korea’s Gwangju Institute of Science and Technology (GIST) have developed a framework for AI-powered in-game observers for esports tournaments.The proposed framework, in an article in the academic journal Expert Systems with Applications, uses an object detection method and human observational data to determine the most interesting regions for the viewer to view.In-game observers are tasked with deciding which players fans should see and which camera angles to use during eSports matches, relying on extensive game knowledge to follow the most important gameplay. If you need more information about South Korea Develop Model then read carefully and share with your friends.
While automatic in-game observers are featured in many games, they usually rely on pre-determined rules and events. According to the researchers, this means that the feature is unable to independently assess the importance of an in-game action.However, researchers at GIST, led by Associate Professor Dr Kyung-Jeong Kim, have proposed an approach that they say can overcome these limitations.
The innovation of the framework lies in treating the field as an object viewed by the viewer, and not as in-game objects such as characters or buildings.In their study, the researchers collected in-game human observation data on StarCraft from 25 participants.Areas seen by the viewer were identified and labeled as ‘one’, while the rest of the screen was filled with ‘zero’.
Also read Inter Milan Teams Up With Italian Esports Organisation Mkers
That data was fed into a neural network to understand patterns between regions, in order to find ‘regions of common interest’ (ROCIs) – essentially, the most exciting regions for viewers to see.The researchers then compared the results with other existing methods, and determined that the network’s predicted approach was similar to human observational data. The paper claims that the model outperforms other models in the long run when testing generalized across different matchups and in-game StarCraft maps.
More importantly, the authors claim the proposed framework “can be applied to a variety of games that can have a game state, such as a minimap, and the area seen by a viewer is part of the game state”. ” League of Legends and Dota 2 were listed as examples.The research could prove valuable, providing a cheaper option for smaller tournament organizers who lack the budget to pay a dedicated observer.
GIST said that should the model prove successful in practice, it could potentially offer an improvement over human observers who may miss important events occurring simultaneously on multiple screens.Study lead and GIST associate professor Dr Kyung-Jong Kim said in a release: “We built an automated supervisor using [an] object detection algorithm, Mask R-CNN, to learn from human observation data.
“The framework can be applied to other sports that represent the overall game state, not just StarCraft. As services such as multi-screen transmission become increasingly common in e-sports, the proposed automated supervisor could be one of these deliverables. Will play a role. It will also be actively used in additional content developed in the future. Along with Kyung-Jong Kim, the paper was co-authored by Ho-Taek Joo, Sung-Ha Lee and Cheong-Mok Bae.Gwangju Institute of Science and Technology is a research-oriented university in Gwangju, a city in the southwest of South Korea.
Get the latest article about esports from Sportslumo.com by following us on Instagram, Twitter, and YouTube
