Detecting Small Chinese Traffic Signs via Improved YOLOv3 Method

Baojun Zhang,Lin Xu,Huilan Wang,Chenchen Xu,Guili Wang,Yu Li,Weijun Zhou

doi:10.1155/2021/8826593

Abstract

Long-distance detection of traffic signs provides drivers with more reaction time, which is an effective technique to reduce the probability of sudden accidents. It is recognized that the imaging size of far traffic signs is decreasing with distance. Such a fact imposes much challenge on long-distance detection. Aiming to enhance the recognition rate of long-distance small targets, we design a four-scale detection structure based on the three-scale detection structure of YOLOv3 network. In order to reduce the occlusion effects of similar objects, NMS is replaced by soft-NMS. In addition, the datasets are trained and the K-Means method is used to generate the appropriate anchor boxes, so as to speed up the network computing. By using these methods, better experimental results for the recognition of long-distance traffic signs have been obtained. The recognition rate is 43.8 frames per second (FPS), and the recognition accuracy is improved to 98.8%, which is much better than the original YOLOv3.

Highlights

IntroductionThe research on intelligent driving has attracted increasing attention
In recent years, the research on intelligent driving has attracted increasing attention
The traffic sign recognition mainly relies on the extraction of color features, shape features, and other methods [2]. e image is segmented according to various colors

Summary

Introduction

The research on intelligent driving has attracted increasing attention. The traffic sign recognition mainly relies on the extraction of color features, shape features, and other methods [2]. Dai et al proposed a new solution to improve the recognition rate of traffic signs in different brightness environments via colors, providing 78% accuracy and 11 FPS, respectively [3]. In [4], Miao used an improved K-Means clustering algorithm to segment color images and used Hough transform to segment traffic signs of different shapes. E accuracy of this method is high, but the FPS is only 4.9. The highest speed is 11 FPS and the lowest is 1.7 FPS. It is obvious that the speed cannot meet the needs of rapid detection, especially for small targets [5]

Methods

Results

Conclusion