Predição de vitória no League of Legends Dois Vizinhos

Abstract

The e-sport market has been growing worldwide, bringing sponsorship from companies from different branches of the economy. Linked to this, more people with little or no contact with electronic games are following their broadcasts. Thereby, a big problem for new viewers is understanding how the game is played since its dynamic is complex and inexperienced viewers may not understand the game scoring system. To improve viewers experience, this work proposes the development of an application that uses image pattern recognition techniques to predict which team has the highest probability of winning the match. The application uses a previously defined prediction model to make the prediction. The model was trained using a database generated by the data extracted from videos of previous matches. As a result, na application was obtained that calculates the probability of winning for each team using the data extracted from a real-time match of the League of Legends. A case study was proposed to evaluate the effectiveness of the application in predicting winning teams, and the results show that an application can, for the most part, correctly predict the winning times with an efficiency rate between 80% and 95 %.