Korean Journal of Artificial Intelligence (한국인공지능학회지)

Korea Artificial Intelligence Association (한국인공지능학회)
권호 표지
DOI QR코드

DOI QR Code

Multi-Scale Dilation Convolution Feature Fusion (MsDC-FF) Technique for CNN-Based Black Ice Detection

  • Sun-Kyoung KANG (Department of Computer Engineering, Wonkwang University)
  • Received : 2023.06.27
  • Accepted : 2023.09.05
  • Published : 2023.09.30
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, we propose a black ice detection system using Convolutional Neural Networks (CNNs). Black ice poses a serious threat to road safety, particularly during winter conditions. To overcome this problem, we introduce a CNN-based architecture for real-time black ice detection with an encoder-decoder network, specifically designed for real-time black ice detection using thermal images. To train the network, we establish a specialized experimental platform to capture thermal images of various black ice formations on diverse road surfaces, including cement and asphalt. This enables us to curate a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Additionally, in order to enhance the accuracy of black ice detection, we propose a multi-scale dilation convolution feature fusion (MsDC-FF) technique. This proposed technique dynamically adjusts the dilation ratios based on the input image's resolution, improving the network's ability to capture fine-grained details. Experimental results demonstrate the superior performance of our proposed network model compared to conventional image segmentation models. Our model achieved an mIoU of 95.93%, while LinkNet achieved an mIoU of 95.39%. Therefore, it is concluded that the proposed model in this paper could offer a promising solution for real-time black ice detection, thereby enhancing road safety during winter conditions.

Keywords

Acknowledgement

This paper was supported by Wonkwang University in 2023.

References

  1. Breckon, T., & Fisher, R. B. (2012). A novel thermal-based approach to black ice detection. In Proceedings of the 21st International Conference on Pattern Recognition (ICPR) (pp. 1492-1495). IEEE.
  2. Chaurasia, A., & Culurciello, E. (2017, December). Linknet: Exploiting encoder representations for efficient semantic segmentation. In 2017 IEEE visual communications and image processing (VCIP) (pp. 1-4). IEEE.
  3. Chen, L. C., Papandreou, G., Schroff, F., & Adam, H. (2017). Rethinking atrous convolution for semantic image segmentation. arXiv preprint arXiv:1706.05587.
  4. Huang, G., Chen, D., Li, T., Wu, F., Van Der Maaten, L., & Weinberger, K. Q. (2017). Multi-scale dense networks for resource efficient image classification. arXiv preprint arXiv:1703.09844.
  5. Kim, S.-J., Yoon, W.-S., & Kim, Y.-K. (2021). Characteristics of Black Ice Using Thermal Imaging Camera. Journal of the Korean Society of Industry Convergence, 24(6_2), 873-882. https://doi.org/10.21289/KSIC.2021.24.6.873
  6. Korea Traffic Accident Analysis System [Internet]. Available: http://taas.koroad.or.kr/.
  7. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). Imagenet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84-90.
  8. Lee, H., Hwang, K., Kang, M., & Song, J. (2020, December). Black ice detection using CNN for the Prevention of Accidents in Automated Vehicle. In 2020 International Conference on Computational Science and Computational Intelligence (CSCI) (pp. 1189-1192). IEEE.
  9. Li, Q., Ji, Y. W., Wang, Z. P., & Dou, X. (2017). Design of Road Icing Detection System Based on Opencv+ Python. Journal of Shaanxi University of Science & Technology (Natural Science Edition), 35(2), 158-164.
  10. Long, J., Shelhamer, E., & Darrell, T. (2015). Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3431-3440).
  11. Park, G. Y., Lee, S. H., Kim, E. J., & Yun, B. Y. (2017). A case study on meteorological analysis of freezing rain and black ice formation on the load at winter. Journal of Environmental Science International, 26(7), 827-836.
  12. Paszke, A., Chaurasia, A., Kim, S., & Culurciello, E. (2016). Enet: A deep neural network architecture for real-time semantic segmentation. arXiv preprint arXiv:1606.02147.
  13. Ronneberger, O., Fischer, P., & Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18 (pp. 234-241). Springer International Publishing.
  14. Smith, S., Williams, B. L., & Prato, C. G. (2017). Black ice detection. In Encyclopedia of Traffic Science (pp. 1-7). Springer.
  15. Wang, Q., Zhang, X., Chen, C., & Li, P. (2019). Black ice detection method based on the temperature field characteristic of thermal images. Journal of Advanced Transportation, 1-14.
  16. Zhao, H., Shi, J., Qi, X., Wang, X., & Jia, J. (2017). Pyramid scene parsing network. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2881-2890).