White Blood Cells type Detection using YOLOv5

White blood cells (WBCs) is the main defense system for human health. It tells about the person health and genetic information, which are able to describe any abnormality occur. A healthy white blood cells are able to identify by its visual characteristic. A common white blood cells disease is the leukemia, which it affect the production of healthy cells. It is crucial to diagnose as soon as possible in order to perform further treatment. Current blood smear diagnoses are done manually by pathologist and it can be very time consuming. This problem can be resolve by developing a computer aided detection system. This paper studied the effect of the latest state-of-art detection model, YOLOv5 on detecting each type of white blood cells (eosinophil, lymphocyte, monocyte, and neutrophil). It is then trained with 2800 images from Kaggle.com public dataset. YOLOv5 have four different model sizes, which are YOLOv5s, YOLOv5m, YOLOv51 and the YOLOv5x. The YOLOv5m achieved the highest mean average precision (mAP) value of 99.42% with 90 minutes of training time.