Artificial Intelligence Based Brain Tumor Localization Using YOLOv5

Abstract

Brain tumor is a mutation in the brain cells in which the cells keep dividing. The earlier the tumor detected, the higher survival rate for the patient. This study develops the brain tumor detection system by utilizing the you only look once (YOLO). The model is based on YOLOv5 architect. The open dataset of tumorous images is utilized. From this dataset, the corresponding masks are given alongside the images. Our study tries to compare several YOLOv5 models to localize the brain tumor. The results show YOLOv5m, YOLOv5l, and YOLOv5x models have higher precision and recall values. The inference time from those models is relatively small for recent computational resources. In conclusion, the YOLOv5 models have produced superior result in localizing the brain tumor