Classification and Analysis of Malaria Infected Blood Cells

Abstract

A contagious illness An viral disease called malaria kills more than 500,000 people annually throughout the world. The majority of these fatalities are brought on by a delayed or inaccurate diagnosis. The manual microscope is now thought to be the best technology for diagnosing malaria. On the other hand, it takes time and is subject to human mistake. It is crucial that the evaluation procedure be automated because it is such a significant issue for world health. This article's goal is to promote automation of the diagnosis procedure in order to do away with the necessity for human involvement. The strategy is founded on known erythrocyte and Plasmodium parasite intensity characteristics to be erratic. The same dataset is used to train the transfer learning models CNN, densnet201, Nasnet large, InceptionNet, Xception, Hybrid (CNN + DenseNet201 + NasNet Large + InceptionNet + Xception), KNN, SVM, Mobilenet, VGG16, Resnet50, InceptionV3, Densenet169, Resnet101, Lenet, and efficientnet V2S. Both transfer learning and fine-tuning strategies are used, and the results are contrasted. Experimental evidence demonstrates that the suggested models operate well and generate reliable outcomes.