FSP > Volume 3, Number 4, October 2019

Multi-Scale CapsNet: A Novel Traffic Sign Recognition Method
DOI: 10.22606/fsp.2019.34005

Author(s)
Gongbin Chen, Yansong Deng
Affiliation(s)
Key Laboratory of Electronic and Information Engineering (Southwest Minzu University), State Ethnic Affairs Commission, Chengdu, China
Key Laboratory of Electronic and Information Engineering (Southwest Minzu University), State Ethnic Affairs Commission, Chengdu, China
Abstract
Convolutional Neural Networks (CNNs) have performed very well on image classification tasks, but CNNs is insensitive to detailed image information and requires a large amount of training data and time. Capsule Networks(CapsNets) can solve this problem very well, but the Baseline CapsNet model is very shallow, and the extraction of low-level features is not enough. We propose a Multi-Scale Capsule Network (Multi-Scale CapsNet), by extracting the low-level features of images with multi-channel convolution of multiple convolution kernels, so extracted features are more diverse, then passing from the bottom layer to the upper layer in the form of a "capsule", which encapsulats the multidimensional features of the image in the form of a vector, thus the features are saved in the network, rather than being recovered after being lost. In the German Traffic Sign Recognition Benchmark(GTSRB), we obtained competitive results with the accuracy of 99.4%, which is better than the human performance of 98.81% and the Multi-Scale Convolutional Neural Network(MS-CNN) of 97.33%.
Keywords
CapsNets, multi-scale CapsNet, traffic sign recognition, multi-scale convolution, CNN
References

• [1]  W. C. U. Ritter and D. B. AG, aAIJTraffic sign recognition in color image sequences,aAI in Intelligent Vehicles ’92 Symposium., Proceedings of the.IEEE, 1992, pp. 12aAS17.
• [2]  G. W. N. Ubong Lydia Jau, Chee Siong Teh, aAIJA comparison of rgb and hsi color segmentation in real - time video images: A preliminary study on road sign detection,aAI in Information Technology, 2008. ITSim 2008. International Symposium on, vol. 4, 2008, pp. 1aAS6.
• [3]  L.-W. Tsai, J.-W. Hsieh, C.-H. Chuang, Y.-J. Tseng, K.-C. Fan, and C.-C. Lee, aAIJRoad sign detection using eigen colour,aAI ComputerVision,IET, vol. 2, no. 3, pp. 164aAS177, 2008.
• [4]  Y. Aoyagi and T. Asakura, aAIJA study on traffic sign recognition in scene image using genetic algorithms and neural networks,aAI in Industrial Electronics, Control, and Instrumentation, 1996., Proceedings of the 1996 IEEE IECON 22nd International Conference on, vol. 3, 1996, pp. 1838aAS1843 vol.3.
• [5]  R. Belaroussi and J. P. Tarel, aAIJAngle vertex and bisector geometric model for triangular road sign detection,aAI in Applications of Computer Vision (WACV), 2009 Workshop on, 2009, pp. 1aAS7.
• [6]  B. Hoferlin and K. Zimmermann, aAIJTowards reliable traffic sign recognition,aAI in Intelligent Vehicles Symposium, 2009 IEEE, 2009, pp. 324aAS329.
• [7]  G. Loy and N. Barnes, aAIJFast shape-based road sign detection for a driver assistance system,aAI in Intelligent Robots and Systems, 2004. (IROS 2004). Proceedings. 2004 IEEE/RSJ International Conference on, vol. 1, 2004, pp. 70aAS75 vol.1.
• [8]  Y.-S. Huang, M.-Y. Fu, and H.-B. Ma, aAIJA combined method for traffic sign detection and classification,aAI in Pattern Recognition (CCPR), 2010 Chinese Conference on, 2010, pp. 1aAS5.
• [9]  P. PaclÅCk, J. Novovicov, P. Pudil, and P. Somol, aAIJRoad sign classification using laplace kernel classifier,aAI Pattern Recognition Letters, pp. 1165aAS1173, 2000.
• [10]  Pierre Sermanet, Yann LeCun. Traffic Sign Recognition with Multi-Scale Convolutional Networks. In International Joint Conference on Neural Networks, 2011.
• [11]  K. Jarrett, K. Kavukcuoglu, M. Ranzato, and Y. LeCun, aAIJWhat is the best multi-stage architecture for object recognition?aAI in Computer Vision, 2009 IEEE 12th International Conference on, 29 2009-oct. 22009, pp. 2146 aAS2153.
• [12]  D. Ciresan, U. Meier, J. Masci, and J. Schmidhuber, aAIJA committee of neural networks for traffic sign classification,aAI in IJCNN’11, 2011, pp. 1918aAS1921.
• [13]  A. Krizhevsky, I. Sutskever, and G. E. Hinton, aAIJImagenet classification with deep convolutional neural networks,aAI in NIPS’12, 2012, pp. 1106aAS1114.
• [14]  SABOUR S, FROSST N, HINTON G E. Dynamic routing between Capsules [J]. NIPS2017, 2017.
• [15]  Geoffrey Hinton, Sara Sabour, and Nicholas Frosst. Matrix capsules with em routing. ICLR 2018, 2018.
• [16]  M. Amer and T. Maul, aAIJPath capsule networks,aAI in preprint, arxiv., 2019.
• [17]  R. Mukhometzianov and J. Carrillo, aAIJCapsnet comparative performance evaluation for image classification, aAI arXiv preprint arXiv:1805.11195, 2018.
• [18]  C. Xiang, L. Zhang, Y. Tang, W. Zou, and C. Xu, aAIJMs-capsnet: A novel multi-scale capsule network,aAI IEEE Signal Processing Letters, vol. 25, no. 12, pp. 1850aAS1854, 2018.
• [19]  Min Lin, Qiang Chen, and Shuicheng Yan. Network in network. CoRR, abs/1312.4400, 2013.