Cell Counting Algorithm Based on YOLOv3 and Image Density Estimation

Abstract

Cell counting is an important branch of medical image processing research. One of the important tasks is to count red cells and white cells in microscopic images. In an image, the distribution of white blood cells is relatively sparse and easy to count, while the distribution of red blood cells is relatively dense, which is prone to overlap adhesion. In this paper, we proposed a new algorithm for counting red blood cells and white blood cells in microscopic images. Firstly, we use YOLOv3 network to detect discrete red cells, white blood cells and aggregated red blood cells in microscopic image. Then we count aggregated red blood cells by image density estimation method. In this way, we can provide more accurate counting results than YOLOv3 and image density estimation method.