Empirical Evaluation of Convolutional Neural Networks Prediction Time in Classifying German Traffic Signs

Joshua Fulco, Akanksha Devkar, Aravind Krishnan, Gregory Slavin, Carlos Morato

2017

Abstract

This paper discusses the use of Deep Learning and neural networks to identify images which contain road signs to aid in the navigation of autonomous vehicles. Images of 32x32 pixels and 128x128 pixels of the GTSRB dataset were used in training the existing neural network models as well as our novel models. Existing neural network models mentioned in the literature study validate that very high accuracies in image classification are already achieved. Different neural network model architectures were also reviewed to determine which architecture produced the highest accuracy within the most efficient time. Modifications to these architectures were made to produce valid results with a reduced image identification time. Our results of classifying a traffic sign image of 32x32 pixels in 0.6ms is very reliable for real time output. By looking at the image identification times for a 32x32 pixel image and a 128x128 pixel image we observed that size of the image is not the main factor in the increase of the prediction time.

References

  1. Bojarski, M., Del Testa, D., Dworakowski, D., Firner, B., Flepp, B., Goyal, P., Jackel, L. D., Monfort, M., Muller, U., and Zhang, J. e. a. (2017). End to end learning for self-driving cars. Arxiv.org.
  2. Chen, L., Li, Q., Li, M., Zhang, L., and Mao, Q. (2012). Design of a multi-sensor cooperation travel environment perception system for autonomous vehicle. Sensors, 12(12):12386-12404.
  3. Ciresan, D., Meier, U., Masci, J., and Schmidhuber, J. (2011). A committee of neural networks for traffic sign classification. The 2011 International Joint Conference on Neural Networks.
  4. Greenhalgh, JackMirmehdi, M. (2012). Real-time detection and recognition of road traffic signs. IEEE Transactions on Intelligent Transportation Systems, 13(4):1498-1506.
  5. Jin, J., Fu, K., and Zhang, C. (2014). Traffic Sign Recognition With Hinge Loss Trained Convolutional Neural Networks. IEEE Transactions on Intelligent Transportation Systems, 15(5):1991-2000.
  6. Le, T. T., Tran, S. T., Mita, S., and Nguyen, T. D. (2010). Real time traffic sign detection using color and shapebased features. Intelligent Information and Database Systems, pages 268-278.
  7. Mao, X., Hijazi, S., Casas, R., Kaul, P., Kumar, R., and Rowen, C. (2016). Hierarchical cnn for traffic sign recognition. 2016 IEEE Intelligent Vehicles Symposium (IV).
  8. Menge, Anheuser-Busch, O. (2016). Complete first selfdriving truck delivery of beer. Neural Networks.
  9. Sermanet, PierreLeCun, Y. (2011). Traffic sign recognition with multi-scale convolutional networks. The 2011 International Joint Conference on Neural Networks.
  10. Stallkamp, J., Schlipsing, M., Salmen, J., and Igel, C. (2012). Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition. Neural Networks, 32:323-332.
  11. Zeng, Y., Xu, X., Fang, Y., and Zhao, K. (2015). Traffic sign recognition using deep convolutional networks and extreme learning machine. Lecture Notes in Computer Science, pages 272-280.
Download


Paper Citation


in Harvard Style

Fulco J., Devkar A., Krishnan A., Slavin G. and Morato C. (2017). Empirical Evaluation of Convolutional Neural Networks Prediction Time in Classifying German Traffic Signs . In Proceedings of the 3rd International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS, ISBN 978-989-758-242-4, pages 260-267. DOI: 10.5220/0006307402600267


in Bibtex Style

@conference{vehits17,
author={Joshua Fulco and Akanksha Devkar and Aravind Krishnan and Gregory Slavin and Carlos Morato},
title={Empirical Evaluation of Convolutional Neural Networks Prediction Time in Classifying German Traffic Signs},
booktitle={Proceedings of the 3rd International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS,},
year={2017},
pages={260-267},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0006307402600267},
isbn={978-989-758-242-4},
}


in EndNote Style

TY - CONF
JO - Proceedings of the 3rd International Conference on Vehicle Technology and Intelligent Transport Systems - Volume 1: VEHITS,
TI - Empirical Evaluation of Convolutional Neural Networks Prediction Time in Classifying German Traffic Signs
SN - 978-989-758-242-4
AU - Fulco J.
AU - Devkar A.
AU - Krishnan A.
AU - Slavin G.
AU - Morato C.
PY - 2017
SP - 260
EP - 267
DO - 10.5220/0006307402600267