Update README.md

README.md

 # 2023-028
 
-SpinPerfom – A low-cost cricket spin bowling performance analysis model
\ No newline at end of file
+**SpinPerfom – A low-cost cricket spin bowling performance analysis model**
+
+
+**Individual component**
+
+To detect accurately that a part of spinner’s foot is always behind the crease (for the bowler to avoid putting a no ball)
+
+**Main objectives** 
+
+Detect whether any part of the front foot is behind the bowling crease to determine the delivery is not a no ball or a legal ball.
+
+**Specific objectives**
+
+To implement whether any part of the front foot is behind the bowling crease to determine the delivery is not a no ball there are 
+three main approaches. Below show what the objectives need when using all the approaches.
+
+•	Collect data
+For the model training purpose, it is required to collect the data. In this individual component two of the following scenarios 
+will be considered.
+                * Moment where any part of the front foot is behind the bowling crease.(legal ball)
+                * Moment where any part of the front foot is not behind the bowling crease.(no ball)
+    
+•	Preprocessing of the data set
+Collect the footages of the players during their practice session and extract the frame.
+Refine the images and organize them into two different folders one as no ball and other as legal ball.
+
+•	Model training and testing
+To detect whether a ball is a no ball or a legal ball using machine learning, do a comparison 
+between Convolutional Neural Network (CNN) model and pretrained model MobileNet, ResNet.
+It can experiment with all three architectures and evaluate their performance on the dataset to select 
+the best model for the goal of detecting no balls, and then provide the best model.
+
+Get the training and accuracy and the accuracy after evaluation
+
+Test the implementation using a test image
+