Multi-column Spatial Transformer Convolution Neural Network for Traffic Sign Recognition

Traffic sign recognition is an important research area in intelligent transportation, which is especially important in autopilot system. Convolutional Neural Network (CNN) is the main research method of traffic sign recognition. However, the convolution neural network is easily affected by the spatial diversity of the image. With regard to this, in this paper, a multi-column spatial transformer convolution neural network named MC-STCNN is proposed to solve the problem when Convolutional Neural Network (CNN) can’t adapt to the spatial diversity of the image very well. The MC-STCNN network consisted of CNN and STN is formed by training pictures of different sizes. It can be well adapted to the spatial diversity and the images input of different sizes. It achieves an accuracy of 99.75% on GTSRB traffic sign recognition, exceeding the current highest accuracy of 99.65%.

[1]  Changshui Zhang,et al.  Traffic Sign Recognition With Hinge Loss Trained Convolutional Neural Networks , 2014, IEEE Transactions on Intelligent Transportation Systems.

[2]  D. L. Kellmeyer,et al.  Detection of highway warning signs in natural video images using color image processing and neural networks , 1994, Proceedings of 1994 IEEE International Conference on Neural Networks (ICNN'94).

[3]  Johannes Stallkamp,et al.  The German Traffic Sign Recognition Benchmark: A multi-class classification competition , 2011, The 2011 International Joint Conference on Neural Networks.

[4]  Jiri Matas,et al.  Systematic evaluation of convolution neural network advances on the Imagenet , 2017, Comput. Vis. Image Underst..

[5]  Andrew Zisserman,et al.  Spatial Transformer Networks , 2015, NIPS.

[6]  Domenec Puig,et al.  A Practical and Highly Optimized Convolutional Neural Network for Classifying Traffic Signs in Real-Time , 2016, International Journal of Computer Vision.

[7]  Reinhard Klette,et al.  General traffic sign recognition by feature matching , 2009, 2009 24th International Conference Image and Vision Computing New Zealand.

[8]  Hidehiko Akatsuka,et al.  Road Signposts Recognition System , 1987 .

[9]  Luc Van Gool,et al.  Integrating Object Detection with 3D Tracking Towards a Better Driver Assistance System , 2010, 2010 20th International Conference on Pattern Recognition.

[10]  Jason Jianjun Gu,et al.  An Efficient Method for Traffic Sign Recognition Based on Extreme Learning Machine , 2017, IEEE Transactions on Cybernetics.

[11]  Fuchun Sun,et al.  Speed Limit Sign Recognition Using Log-Polar Mapping and Visual Codebook , 2012, ISNN.

[12]  Jürgen Schmidhuber,et al.  Multi-column deep neural network for traffic sign classification , 2012, Neural Networks.