Overview of YOLO Object Detection Algorithm

Abstract

As an important research direction in the field of computer vision, object detection has developed rapidly and many kinds of mature algorithms emerged. The series of YOLO (You Only Look Once) algorithms implement one-stage detection based on regression ideas, which showing preeminent in speed and owning strong generalization on a variety of datasets. This paper will give a simple introduction to the current mainstream deep learning object detection algorithm, then focus on combing the principle and optimizational process of the series of YOLO algorithms, summarize the latest breakthroughs in YOLO algorithm, Hopefully that can provide reference for the research of related topics.