Analysis of Overhead View Images at Intersection Using Machine Learning

  • Taisuke Hori
  • Mitsuhiro NamekawaEmail author
  • Syuya Kanagawa
Conference paper
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 940)


In recent years, cameras are developing smaller and higher performance. Therefore, it became possible to take images from free position. In this research, we aim to analyze the traffic flow by take the overhead view image from the top of the building. We have constructed a system to accurately track the position of vehicles using image recognition technology with machine learning. In order to construct this system, images taken at three different intersection points were prepared as machine learning teacher images. This makes it possible to automatically measure the running time of the vehicle and the group of vehicles at the intersection.

As result of this research, we can clarify the features and problems of intersections and roads. And the future, we will construct the vehicle simulation system using this result, highly accurate congestion analysis and congestion prediction become possible.


Image recognition Machine learning Traffic simulation 


  1. 1.
    Takahashi, M., Namekawa, M., Satoh, A.: An analysis of road traffic data in a congestion condition for MITRAM. In: Proceedings of the 11th Simulation Technology Conference, JSST (1992)Google Scholar
  2. 2.
    Feldkamp, N., Strassburger, S.: Automatic generation of route networks for microscopic traffic simulations. In: Proceedings of the 2014 Winter Simulation Conference, pp. 2848–2859. IEEE Press (2014)Google Scholar
  3. 3.
    Takahashi, R., et al.: Explaining potential risks in traffic scenes by combining logical inference and physical simulation. Int. J. Mach. Learn. Comput. 6(5), 248 (2016)CrossRefGoogle Scholar
  4. 4.
    Teknomo, K.: Microscopic pedestrian flow characteristics: development of an image processing data collection and simulation model. Ph.D. dissertation at Department of Human Social Information Sciences, Graduate School of Information Sciences, Tohoku University, Japan (2002). arXiv preprint arXiv:1610.00029 (2016)
  5. 5.
    Ide, T., et al.: City-wide traffic flow estimation from limited number of low quality cameras. IEEE Trans. Intell. Transp. Syst. 18(4), 950–959 (2017)CrossRefGoogle Scholar
  6. 6.
    Fujimoto, R., et al.: A dynamic data driven application system for vehicle tracking. Procedia Comput. Sci. 29, 1203–1215 (2014)CrossRefGoogle Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  • Taisuke Hori
    • 1
  • Mitsuhiro Namekawa
    • 1
    Email author
  • Syuya Kanagawa
    • 2
  1. 1.Kaetsu UniversityTokyoJapan
  2. 2.Tokyo City UniversityTokyoJapan

Personalised recommendations