Advertisement

Traffic Sign Detection Using R-CNN

  • Philipp RehlaenderEmail author
  • Maik Schroeer
  • Gavneet Chadha
  • Andreas Schwung
Conference paper
Part of the Proceedings of the International Neural Networks Society book series (INNS, volume 1)

Abstract

Due to the increasing popularity of driving assistance systems, traffic sign detection has gained increased attention due to its role in traffic safety. This paper presents a traffic sign detection algorithm based on the regions with convolutional neural network (R-CNN) method. In this method, a segmentation algorithm proposes regions likely to contain a traffic sign. These regions are analyzed by a convolutional neural network providing a feature vector to a support vector machine. This method, however, suffers from a high computation time due to an excessive amount of regions proposed. To reduce the number of regions, this work proposes a perspective based logic that discards regions unlikely to contain traffic signs because of a size-position-mismatch. This results in an average 93% reduction of proposed regions. For faster training, the system consists of a two-layer support vector machine. The first layer is used to categorize the family of the picture, the second set of SVMs for classifying the traffic sign. The overall classification accuracy of this technique is 86%. Feeding the results to a tracking algorithm may result in an overall improvement because frames would compensate errors.

Keywords

Object detection Convolutional neural network Regions with convolutional neural networks (R-CNN) Support vector machines (SVM) 

References

  1. 1.
    Boi, F., Gagliardini, L.: A support vector machines network for traffic sign recognition. In: The 2011 International Joint Conference on Neural Networks, pp. 2210–2216. IEEE (2011).  https://doi.org/10.1109/IJCNN.2011.6033503
  2. 2.
    Girshick, R., Donahue, J., Darrell, T., Malik, J.: Rich feature hierarchies for accurate object detection and semantic segmentation. In: 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 580–587. IEEE (2014).  https://doi.org/10.1109/CVPR.2014.81
  3. 3.
    Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep convolutional neural networks. http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks
  4. 4.
    Maldonado-Bascon, S., Lafuente-Arroyo, S., Gil-Jimenez, P., Gomez-Moreno, H., Lopez-Ferreras, F.: Road-sign detection and recognition based on support vector machines. IEEE Trans. Intell. Transp. Syst. 8(2), 264–278 (2007).  https://doi.org/10.1109/TITS.2007.895311CrossRefzbMATHGoogle Scholar
  5. 5.
    Mao, X., Hijazi, S., Casas, R., Kaul, P., Kumar, R., Rowen, C.: Hierarchical CNN for traffic sign recognition. In: 2016 IEEE Intelligent Vehicles Symposium (IV), pp. 130–135. IEEE (2016).  https://doi.org/10.1109/IVS.2016.7535376
  6. 6.
    Qian, R.Q., Yue, Y., Coenen, F., Zhang, B.L.: Traffic sign recognition using visual attribute learning and convolutional neural network. In: 2016 International Conference on Machine Learning and Cybernetics (ICMLC), pp. 386–391. IEEE (2016).  https://doi.org/10.1109/ICMLC.2016.7860932
  7. 7.
    Shi, M., Wu, H., Fleyeh, H.: Support vector machines for traffic signs recognition. In: 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence), pp. 3820–3827. IEEE (2008).  https://doi.org/10.1109/IJCNN.2008.4634347
  8. 8.
    Stallkamp, J., Schlipsing, M., Salmen, J., Igel, C.: Man vs. computer: benchmarking machine learning algorithms for traffic sign recognition. Neural Networks Off. J. Int. Neural Network Soc. 32, 323–332 (2012).  https://doi.org/10.1016/j.neunet.2012.02.016CrossRefGoogle Scholar
  9. 9.
    Uijlings, J.R.R., van de Sande, K.E.A., Gevers, T., Smeulders, A.W.M.: Selective search for object recognition. Int. J. Comput. Vis. 104(2), 154–171 (2013).  https://doi.org/10.1007/s11263-013-0620-5CrossRefGoogle Scholar
  10. 10.
    Vedaldi, A., Lenc, K.: MatConvNet: convolutional neural networks for MATLAB (2014). https://arxiv.org/pdf/1412.4564v3.pdf

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  • Philipp Rehlaender
    • 1
    Email author
  • Maik Schroeer
    • 2
  • Gavneet Chadha
    • 3
  • Andreas Schwung
    • 3
  1. 1.Paderborn UniversityPaderbornGermany
  2. 2.Busch-Jaeger Elektro GmbHLüdenscheidGermany
  3. 3.South Westphalia University of Applied SciencesIserlohnGermany

Personalised recommendations