Abstract : Cricket is one of the most popular sports that is played in many countries and have audience who like it in huge numbers. And in the technology field, Artificial Intelligence has gained lots of interest from people. So, implementing Artificial Intelligence in the field of sports, especially cricket, has brought many advancements which has helped in decision making, ball tracking and many visualizations. In this project we have implemented a model which classifies the cricket shots using the pose of the player and also based on the shot obtained we predict the runs that can be scored. As per the survey many algorithms and techniques have been proposed which include Artificial neural network, Computer vision, Deep convolutional neural network, Pose detection, Long short term memory, Recurrent Neural Network and Deep neural network. The Project has been implemented using Convolution neural network based algorithms which can be used for self-paced training and to predict score from a shot. In other words, the model identifies the shot of the player which can be related to the pose structure, by which the players can improve their shot pose, helping in training themselves. The dataset has been obtained from Kaggle where each shot has around 1000 images, with a total of around 4000 images. The results are obtained from the CNN models that are VGG-16 and ResNet-50, where the better results are obtained using ResNet-50. The dataset has been divided into 80% and 20% for training and testing purposes respectively. The score prediction for shot classification is done using the Linear regression model. This can help batsmen to improve their shot, bowlers to ball so that they can take wickets or reduce the runs and also would be helpful in training the players.
Keywords : Cricket shot, Convolution neural network, Resnet-50, VGG-16, Linear regression.
|
DOI:
10.17148/IJARCCE.2023.125178
