Research on Mask RCNN based on rotating self-supervised learning
Xuedong Wang, Shi Chen Su, Hongcheng Huang, Pengzhi Chu · 2022 41st Chinese Control Conference (CCC) · 2022
Traffic sign perception is an important part of the field of autonomous driving perception, at present stage, sensors such as cameras, ultrasonic radar, and lidar are usually used to recognize traffic signs. The task of object detection based on deep learning is to find objects of interest in images or videos, and to detect their positions and sizes at the same time. The field of supervised learning is booming in recent years, some researchers have combined self-supervised learning with some downstream tasks in computer vision (such as classification, object detection, semantic segmentation, etc.), and achieved good results. In this article, we design a Mask RCNN object detection network model combined with rotating self-supervised learning, and call it Rot-Mask RCNN, then train the model on the data set that collected by the Hilens Kit camera, and finally test its accuracy. The mAP value can reach 0.9, which fully proves the effectiveness of the method. Data set will be made public later.