Page 137 - Proceedings of the 2018 ITU Kaleidoscope

Machine learning for a 5G future

Dynamic thresholding in surveillance system will make the [9] W. Li, V. Mahadevan, and N. Vasconcelos,
model more real time impermeable. The proposed work can “Anomaly detection and localization in crowded
be further extended by implementing clustering part with scenes,” IEEE Trans. on Pattern Analysis Matching
hierarchical algorithms instead of basic k-means. The Intelligence, vol. 36, no. 1, pp. 18–32, Jan. 2014.
advantage of using hierarchical algorithms is that the
cluster size need not be pre-determined in most cases. The [10] V. Saligrama and Z. Chen, “Video anomaly
proposed method deals only with static cameras. However, detection based on local statistical aggregates,” in
it can be extended to tilt and zoom cameras using Proc. IEEE Conf. Computer Visual Pattern
localization results. Further, camera motion estimation can Recognition (CVPR), pp. 2112–2119, Jun. 2012.
be added to reduce inaccuracies due to small camera
movements. [11] J. Shao, C. C. Loy, and X. Wang, “Scene-
independent group profiling in crowd,” in Proc.
REFERENCES IEEE Conf. Computer Visual Pattern Recognition
(CVPR), pp. 2227–2234, Jun. 2014.
[1] D. Helbing and P. Molnar, “Social force model for
pedestrian dynamics,” Physical Review E, vol. 51, [12] F. Jiang, J. Yuan, S. A. Tsaftaris, and A. K.
no. 5, pp. 4282–4286, 1995. Katsaggelos, “Anomalous video event detection
using spatiotemporal context,” Computer Visual
[2] R. Mehran, A. Oyama, and M. Shah, “Abnormal Image Understand., vol. 115, no. 3, pp. 323–333,
crowd behavior detection using social force model,” 2011.
in Proc. IEEE Conf. Computer Visual Pattern
Recognition, Miami, FL, pp. 935–942, Jun. 2009. [13] Aravinda S. Rao, Jayavardhana Gubbi, Slaven
Marusic, and Marimuthu Palaniswami, “Crowd
[3] X. Cui, Q. Liu, M. Gao, and D. N. Metaxas, Event Detection on Optical Flow Manifolds”, IEEE
“Abnormal detection using interaction energy Transactions on Cybernetics, vol. 46, no. 7, July
potentials,” in Proc. IEEE Conf. Computer Visual 2016
Pattern Recognition, Colorado Springs, CO, USA,
pp. 3161–3167, Jun. 2011. [14] Arnold Wiliem, VamsiMadasu, Wageeh Boles,
Prasad Yarlagadda, “A suspicious behaviour
[4] D. Xu, E. Ricci, Y. Yan, J. Song, and N. Sebe, detection using a context space model for smart
“Learning deep representations of appearance and surveillance systems,” in Computer Vision and
motion for anomalous event detection,” in Proc. Image Understanding, Vol. 116, no. 2, pp. 194 –
British Machine Vision Conference (BMVC), UK, 209, Feb 2012.
pp. 548–561, September 2015.
[15] T. Xiang, S. Gong, “Incremental and adaptive
[5] L. Kratz and K. Nishino, “Anomaly detection in abnormal behaviour detection,” Journal on
extremely crowded scenes using spatio-temporal Computer Vision and Image Understanding, vol.
motion pattern models,” in Proc. IEEE Conf. 111, no. 1, pp. 59–73, July 2008.
Computer Visual Pattern Recognition (CVPR), pp.
1446–1453, Jun 2009. [16] Y. Yuan, J. Fang, and Q. Wang, “Online anomaly
detection in crowd scenes via structure analysis,”
[6] E. L. Andrade, S. Blunsden, and R. B. Fisher, IEEE Trans. Cybern., vol. 45, no. 3, pp. 548–561,
“Modelling crowd scenes for event detection,” in March 2015.
th
Proc. 18 Int. Conf. Pattern Recognition (ICPR), pp.
175–178, August 2006. [17] R. V. H. M. Colque, C. A. C. Júnior, and W. R.
Schwartz, “Histograms of optical flow orientation
[7] J. Kim and K. Grauman, “Observe locally, infer and magnitude and entropy to detect anomalous
globally: A space-time MRF for detecting abnormal events in videos,” IEEE Transactions on Circuits
activities with incremental updates,” in Proc. IEEE and Systems for Video Technology, vol. 27, no. 3,
Conf. Computer Visual Pattern Recognition, Miami, March 2017.
FL, USA, pp. 2921–2928, Jun. 2009.
[18] R. V. H. M. Colque, C. A. C. Júnior, and W. R.
[8] K. W. Cheng, Y. T. Chen, and W. H. Fang, “Video Schwartz, “Histograms of optical flow orientation
anomaly detection and localization using and magnitude to detect anomalous events in videos,”
hierarchical feature representation and Gaussian in Proc. 28th SIBGRAPI Conf. Graph., Patterns
process regression,” in Proc. IEEE Conf. Computer Images, Salvador, Brazil, pp. 126–133, Aug. 2015.
Visual Pattern Recognition (CVPR), pp. 2909–2917,
June 2015. [19] Semiotics Visual Communication Lab UCSD
Anomaly Data Set. [Online]. Available:
http://www.svcl.ucsd.edu/projects/anomaly/

– 121 –