Classification of Degraded Traffic Signs Using Flexible Mixture Model and Transfer Learning

Abdul Mannan,Kashif Javed,Haroon A Babri,Atta Ur Rehman,Serosh K Noon

doi:10.1109/access.2019.2947069

Abstract

Automatic detection and recognition of traffic signs is a topic of research for various applications like driver assistance, inventory management and autonomous driving. Poorly maintained traffic signs degrade by losing their colors or some part is weird due to aging and hence making the task more challenging. The problem is mainly related to the developing world and has gained less attention in the literature on automatic traffic sign detection and recognition. To handle the degradation issue, we present a novel flexible Gaussian mixture model based technique with automatic split and merge strategy. This adaptive scheme works as a preprocessing step which facilitates locating traffic signs under a certain degree of degradation in a real world scenario. A multiscale convolutional neural network augmented with dimensionality reduction layer is proposed to recognize contents of the sign. Since, there is no available benchmark dataset for this purpose, we collected a number of images containing partially degraded signs from the famous German Traffic Sign Detection Benchmark (GTSDB) and augmented it with manually and naturally degraded traffic sign images taken from the longest highway in the country of authors’ residence. Experimental results show that our proposed technique outperforms many state of the art and recent methods for detection and recognition of degraded traffic signs.

Highlights

Traffic signs are placed to convey important road and road-side information to drivers e.g., the recommended speed or warning about a pedestrian or school etc
Rest of this paper is organized as follows: Section II describes work related to degraded traffic sign and convolutional neural networks, Section III explains at length our proposed method for detection and recognition of degraded traffic signs, in Section IV, we provide experimental settings and results and comparison with other state of the art and recent methods, Section V demonstrates conclusion and future work
We find many efforts to detect and recognize standard traffic signs using convolutional neural network (CNN) [1], [19] but their use for degraded signs is yet to be explored

Summary

INTRODUCTION

Traffic signs are placed to convey important road and road-side information to drivers e.g., the recommended speed or warning about a pedestrian or school etc. [1]–[5] (2) using statistical parameter estimation for colors of interest [6]–[8] (3) using saliency and/or dictionary learning to segment windows containing a traffic sign with a certain confidence level [9], [10] and (4) using convolutional neural networks (CNN) to indicate that a real world image contains a traffic sign [11]–[13]. CNN based methods are useful even for recognition purposes but since a lot of parameters are to be learned, they require huge training data and need specialized computing hardware [17], [18]. Rest of this paper is organized as follows: Section II describes work related to degraded traffic sign and convolutional neural networks, Section III explains at length our proposed method for detection and recognition of degraded traffic signs, in Section IV, we provide experimental settings and results and comparison with other state of the art and recent methods, Section V demonstrates conclusion and future work

RELATED WORK

EXPERIMENTS AND RESULTS

CONCLUSION