We have developed a SoftMax based convolutional neural network and a SVM based convolutional neural network for classification of cricket shots. Later, we have comapred the performance of both models using different metrics. There was an existing shotnet model whose accuracy was below 80%. We have upgraded that existing model with completely different techniques, which gives us an accuracy of 92.2%(more than 90% on an average).

Link for Codes:

CNN-SoftMax Model: https://colab.research.google.com/drive/1gG0sHzLIVuv-MlsBUn26N9c-cLwyk2qp?usp=sharing CNN-SVM Model: https://colab.research.google.com/drive/1aJa6GBzCBWQGz_H7uDgrFQL8iLYJOmkW?usp=sharing

Link to folder: https://drive.google.com/drive/folders/1UzvybenTRZ8inpVgouJdLaxtCOnhvvbU?usp=sharing

How to run the code

1. Copy the folder (https://drive.google.com/drive/folders/1UzvybenTRZ8inpVgouJdLaxtCOnhvvbU?usp=sharing) to your drive.
2. Mount the drive and change the 'path_img' variable in the second code cell according to the location of the copied folder in your drive.
3. Run the code cells one by one and it will give the required outputs.

NOTE: We have commented the dataset reading part because it is time consuming and we have saved the dataset as numpy array already in the folder.