In the last 20 years, approximately 7.6 million people have died from the mosquito-borne disease Plasmodium malaria. Time-consuming and difficult to diagnose, standard detection methods provide serious obstacles. So, this study checks how well a CNN model called ShuffleNet and a convolutional neural network (CNN) that uses transfer learning can tell the difference between malaria parasite cells in thick blood smears that are infected and those that are not. The total number of photos is 26,161, with an equal distribution between infected and non-infected cell categories. The study utilized this dataset. The suggested approach shows impressive performance, with training accuracy of 98.77% and testing accuracy of 1.09%. These findings show how well the model performs in classifying fresh data and demonstrate its great stability, which could lead to advancements in machine learning for malaria detection.