Detection of Different Types of Blood Cells: A Comparative Analysis

Object detection is known as the pinnacle of Artificial Intelligence. Object detection is utilized in various fields like healthcare, agriculture, sports, manufacturing, security, and many more. Object detection has numerous different techniques which provide efficient outcomes in different ways. Some algorithms provide better accuracy whereas some are faster than others. In this research, four major image processing algorithms are compared: You Only Look Once, Single Shot Multibox Detector, Faster Region-Based Convolutional Network, and Region-Based Fully Convolutional Network. The comparison of these algorithms is done using the Blood Cells Dataset of Microsoft (Common Object in Context) based on speed and accuracy. The dataset consists of three classes, namely, Platelets, Red Blood Cells, and White Blood Cells. Altogether, the dataset has 364 images and 4888 labels. After completion of the comparative analysis of these four algorithms, it is analyzed that yolo_v3 outperforms all other models in terms of both speed and accuracy.