Machine learning algorithms self-adjust internal parameters such as weights by learning from data. These kinds of parameters are referred to as model parameters. Model parameters are those that are not manually set during the training of the model. Hyper parameters are the parameters that will remain constant and are manually set before the learning model begins to classify the model that is distinct from model parameters. Hyper parameter adjustment is typically necessary while developing a deep learning model. To find the ideal set of hyper parameters, image segmentation techniques like grid search, random search, and Bayesian optimization are frequently employed in image segmentation for tuning hyper parameters of deep convolution networks. Choosing correct hyper parameters or configuring the hyper parameters correctly will enable the learning model to produce accurate predictions. Learning rate, batch size, and kernel size selection are examples of hyper parameters.