A Deep Learning Approach to Predict Batting Strikes Played by a Batsman for Different Bowling Deliveries

In Cricket, players and coaches employ various techniques to win matches. They anticipate the opposing team’s batsmans’ performance in various scenarios and prepare their bowlers appropriately. Similarly, they anticipate the strengths and weaknesses of their teams’ batters against the opposing team’s bowlers and advise them to try various techniques in certain situations. This study proposes a novel CNN model combined with a GRU network to predict batter’s strikes for different bowling deliveries in Cricket, using videos of previously played matches. The proposed model was tested using a new dataset of 428 bowling delivery videos bowled to an English batsman, Joeseph Root, during test Cricket matches. On the prepared dataset, the proposed model attained an overall accuracy of 91.67, classifying four different batting strikes: the Cut shot, the Drive shot, the Flick shot, and the Sweep shot.