A Method of Detect Traffic Police in Complex Scenes

Target detection has a wide range of applications in many areas of life, and it is also a research hotspot in the field of unmanned driving. Urban roads are complex and changeable, especially at intersections, which have always been a difficult and key part in the research of pilotless technology. Traffic policemen detection at intersections is a key link, but there are few existing algorithms, and the detection speed is generally slow. Aiming at this problem, this paper proposes a real-time detection method of traffic police based on YOLOv3 network.The YOLO network is robust and capable of quickly completing target detection tasks. According to the information investigated, there are currently few data sets on traffic police detection. In response to this problem, this paper adopts the transfer learning method, adopts the imageNet set to training model, learns the basic characteristics of people, and then selects 1000 pictures containing traffic police to conduct experiments. The average accuracy of traffic police detection is 77%, and the detection speed reaches 50FPS, which basically meets the requirements of real-time performance, indicating that the method is reasonable and feasible.

[1]  Jianming Zhang,et al.  A Real-Time Chinese Traffic Sign Detection Algorithm Based on Modified YOLOv2 , 2017, Algorithms.

[2]  Ali Farhadi,et al.  YOLO9000: Better, Faster, Stronger , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[3]  Wei Liu,et al.  SSD: Single Shot MultiBox Detector , 2015, ECCV.

[4]  Li Keqiang,et al.  State-of-the-art and technical trends of intelligent and connected vehicles , 2017 .

[5]  William R. Schwartz,et al.  A Robust Real-Time Automatic License Plate Recognition Based on the YOLO Detector , 2018, 2018 International Joint Conference on Neural Networks (IJCNN).

[6]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[7]  Xiaogang Wang,et al.  Pyramid Scene Parsing Network , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Hongyuan Wang,et al.  A vehicle real-time detection algorithm based on YOLOv2 framework , 2018, Commercial + Scientific Sensing and Imaging.

[9]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[10]  Shaoyi Du,et al.  A review of object detection based on deep learning , 2020, Multimedia Tools and Applications.

[11]  Xiaodong Yang,et al.  Video you only look once: Overall temporal convolutions for action recognition , 2018, J. Vis. Commun. Image Represent..

[12]  Ali Farhadi,et al.  YOLOv3: An Incremental Improvement , 2018, ArXiv.

[13]  Ali Farhadi,et al.  You Only Look Once: Unified, Real-Time Object Detection , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).