A Novel Learning-to-Rank Method for Automated Camera Movement Control in E-Sports Spectating

  • Hendi Lie
  • Darren LukasEmail author
  • Jonathan Liebig
  • Richi Nayak
Conference paper
Part of the Communications in Computer and Information Science book series (CCIS, volume 996)


The popularity of modern competitive gaming (or e-sports) has skyrocketed in the past decade. A key part of e-sports is the spectating experience where fans watch tournament games through a camera of the observer. Bigger tournaments hire professional human observers with high-end tools to monitor important events in the game map for broadcasting the game. This setup is prone to errors. It results in missing important events within the game and lowers the spectating experience overall. It is also not sustainable in long-term and not affordable for the small-scale tournaments. This paper proposes a novel method of automated camera movement control using the AdaRank learning-to-rank algorithm to find and predict important events so the camera can be focused on time. The Dota 2 game setup and its replay data are used in extensive experimental testing. The proposed method has shown to outperform the accuracy of both a past machine learning approach and a professional team of human observers.


E-sports Learning-to-rank Automated camera control Dota 2 


  1. 1.
    Khan, I.: Dota 2’s The International 7 breaks esports prize pool record. Accessed 15 Aug 2018
  2. 2.
    Perez, M.: Report: Esports to Grow Substantially and Near Billion-Dollar Revenues in 2018. Accessed 15 Aug 2018
  3. 3.
    Hamari, J., Sjöblom, M.: What is eSports and why do people watch it? Internet Res. 27, 211–232 (2017)CrossRefGoogle Scholar
  4. 4.
    Cook, M., Summerville, A., Colton, S.: Off The Beaten Lane: AI Challenges In MOBAs Beyond Player Control. arXiv Artificial Intelligence (cs.AI) (2017)Google Scholar
  5. 5.
    Phang, D.W.: Intelligent Camera Control in Game Replays (2014)Google Scholar
  6. 6.
    Did you know that the directed camera is broken?: DotA2. Accessed 15 Aug 2018
  7. 7.
    Xu, J., Li, H.: AdaRank: a boosting algorithm for information retrieval. In: Proceedings of the 30th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 391–398. ACM Press, New York (2007)Google Scholar
  8. 8.
    Valve Inc.: Dota 2 (2018).
  9. 9.
    OpenDota: OpenDota API. Accessed 15 Aug 2018
  10. 10.
    Semenov, A., Romov, P., Neklyudov, K., Yashkov, D., Kireev, D.: Applications of machine learning in Dota 2: Literature review and practical knowledge sharing. In: CEUR Workshop Proceedings, pp. 1–5 (2016)Google Scholar
  11. 11.
  12. 12.
    Vinyals, O., Gaffney, S., Ewalds, T.: DeepMind and Blizzard open StarCraft II as an AI research environment.
  13. 13.
    Tamir, M., Oz, G., Ridnik, T.: Method and system for automatic television production (2017)Google Scholar
  14. 14.
    Fandom: Spectator Mode - League of Legends WikiGoogle Scholar
  15. 15.
    Dota 2 Gamepedia: Spectating - Dota 2 WikiGoogle Scholar
  16. 16.
    Liu, T.-Y.: Learning to rank for information retrieval. Found. Trends Inf. Retrieval 3, 225–331 (2007)CrossRefGoogle Scholar
  17. 17.
    Li, H.: A short introduction to learning to rank. IEICE Trans. Inf. Syst. E94-D, 1854–1862 (2011)CrossRefGoogle Scholar
  18. 18.
    Niu, S., Guo, J., Lan, Y., Cheng, X.: Top-k learning to rank: labeling, ranking and evaluation. In: Proceedings of the 35th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 751–760 (2012)Google Scholar
  19. 19.
    Wang, Y., Wang, L., Li, Y., He, D., Liu, T.-Y., Chen, W.: A theoretical analysis of NDCG type ranking measures. In: Conference on Learning Theory, pp. 1–26 (2013)Google Scholar
  20. 20.
    Schrodt, M.: Clarity - Comically fast Dota 2 and CSGO replay parser (2018).
  21. 21.

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  • Hendi Lie
    • 1
  • Darren Lukas
    • 1
    Email author
  • Jonathan Liebig
    • 2
  • Richi Nayak
    • 1
  1. 1.Queensland University of TechnologyBrisbaneAustralia
  2. 2.Liebig ProductionsAachenGermany

Personalised recommendations