Authors:
Yan Lai
;
Nanxin Wang
;
Yusi Yang
and
Lan Lin
Affiliation:
Tongji University, China
Keyword(s):
Traffic Signs Recognition, Convolutional Neural Network, YCbCr Color Space, Support Vector Machine.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Cardiovascular Imaging and Cardiography
;
Cardiovascular Technologies
;
Computer Vision, Visualization and Computer Graphics
;
Geometry and Modeling
;
Health Engineering and Technology Applications
;
Image-Based Modeling
;
Pattern Recognition
;
Signal Processing
;
Software Engineering
Abstract:
Traffic signs recognition and classification play an important role in the unmanned automatic driving. Various
methods were proposed in the past years to deal with this problem, yet the performance of these algorithms
still needs to be improved to meet the requirements in real applications. In this paper, a novel traffic signs
recognition and classification method is presented based on Convolutional Neural Network and Support Vector
Machine (CNN-SVM). In this method, the YCbCr color space is introduced in CNN to divide the color
channels for feature extraction. A SVM classifier is used for classification based on the extracted features.
The experiments are conducted on a real world data set with images and videos captured from ordinary car
driving. The experimental results show that compared with the state-of-the-art methods, our method achieves
the best performance on traffic signs recognition and classification, with a highest 98.6% accuracy rate.