State-of-the-art of Object Detection Model Based on YOLO

Abstract

Object detection has got rapid development these years, and target classification and location become gradually mature. Fast R-CNN, which is a landmark in accuracy, is a kind of classic two-stage model and has relatively low speed. To improve the detection efficiency, YOLO (You Only Look Once) model is proposed. In this article, the research starts from the evolution of CNN for object detection and YOLO model is sorted out. Then we compare the structures, target outputs and network training of each versions. The functional layers evolved into a structure constructed with residual modules, which keep the accuracy. At last, we discuss the system advantage future develop trend. Form the discussion, we conclude that YOLO learns very general representations of objects, and it integrates various mature new technologies. With the improvement of hardware technology, it will be applied more widely.