Detection of Cricket Shots Using Deep Learning

Classifying the different types of bats hots played in cricket has always been a challenging task in the field of cricket indexing .Identifying the type of shot bats man played during a match is a crucial aspect that has not been thoroughly studied. This information can be used for context-based advertisements for cricket viewers, creating sensor-based commentary systems, and coaching assistants. However, manually identifying the different hots from video frames is difficult due to the similarity between them. This project presents a new approach for recognizing and categorizing different crickets hots by using state-of-the-art techniques such as saliency and optical flow to capture both static and dynamic information, and Long Short Term Memory (LSTM) for representation extraction. Additionally, a new data set of120 videos has been introduced to evaluate the performance of the model, with 4 classes of shot search having 30videos.Themodelachievedanaccuracyof83.34%for the four classes of crickets’ hots. Key Word: Cricket, Convolution Neural Network, LSTM, Media Pipe