TR2018-188

Street Scene: A new dataset and evaluation protocol for video anomaly detection


    •  Jones, M.J., Ramachandra, B., "Street Scene: A new dataset and evaluation protocol for video anomaly detection", arXiv, January 2019.
      BibTeX Download PDF
      • @article{Jones2018jan,
      • author = {Jones, Michael J. and Ramachandra, Bharathkumar},
      • title = {Street Scene: A new dataset and evaluation protocol for video anomaly detection},
      • journal = {arXiv},
      • year = 2018,
      • month = jan,
      • url = {https://www.merl.com/publications/TR2018-188}
      • }
  • MERL Contact:
  • Research Areas:

    Artificial Intelligence, Computer Vision, Machine Learning


Progress in video anomaly detection research is currently slowed by small datasets that lack a wide variety of activities as well as flawed evaluation criteria. This paper aims to help move this research effort forward by introducing a large and varied new dataset called Street Scene, as well as two new evaluation criteria that provide a better estimate of how an algorithm will perform in practice. In addition to the new dataset and evaluation criteria, we present two variations of a novel baseline video anomaly detection algorithm and show they are much more accurate on Street Scene than two state-of-the-art algorithms from the literature.