Main Article Content

Abstract

The aim of the project is detecting abnormalities in the crowd behavior.In terms of crowd behavior, while people walks in slow place, if they suddenly started to run then the alarm rings.The propose system first detect the optical flow between frames by using two frame polynomial expansion co-efficients.The magnitude of the difference of two consecutive frames will give us the activity map. The activity map shows how much motion is estimated between consecutive frames. The second by using the differences between activity maps, we obtain Temporal Occupancy Variation (TOV) level and Entropy level.

Article Details

How to Cite
Y.Dhushara, P.Kaviranjani, S.S.Kovarrthina, & V.Gokul brindha. (2018). Deep-cascade: Cascading 3D Deep Neural Networks for Fast Anomaly Detection and Localization in Crowded Scenes . International Journal of Intellectual Advancements and Research in Engineering Computations, 6(2), 1947–1954. Retrieved from https://ijiarec.com/ijiarec/article/view/761