A COMPARATIVE EVALUATION OF DEEP LEARNING METHODS IN DIGITAL IMAGE CLASSIFICATION

Abstract

White Blood Cells are important in determining a person's overall health. The blood disease diagnosis includes characterization and identification of blood samples of a patient. Neural Networks (NN), Convolutional Neural Networks (CNN), and a mix of CNN and NN models are used in recent techniques to improve visual content understanding. From start to finish, The authors were driven to uncover remarkable characteristics in example photographs because of their expertise in medical image analysis. For blood cell classification, the overall performance of individual cell patches extracted using blood smear techniques has been excellent. These approaches, on the other hand, are incapable of dealing with the issue of multiple cells overlapping. Because of the blood cell overlapping pictures, the input image dimension is compressed, the classification time is reduced, as well as the network works better with more accurate parameter estimates. In this review, we are evaluating a detailed scientific comparison of some of the ways used to improve WBC classification. The authors will show some of the ways used to automatically classify their cells. The results of some of the tests used using available data, compared to blood cell classification techniques.