Data-Science-Regular-Bootcamp
Data-Science-Regular-Bootcamp copied to clipboard
3DCNN for anomaly detection
Identify the instant where the offender appears for the first time in the video. Detect the moment when the offender undoubtedly commits a crime. This moment is referred to as the Strict Crime Moment (SCM). 3DCNN is a recent approach for spatio-temporal analysis that has shown remarkable performance in processing videos in different areas, such as moving objects action recognition precision and recall values of 0.857 Reason of the paper : This behavior analysis is the principal reason why we remove the committed crime segment from the video samples, to allow the artificial model to focus on decisive conduct and not in the offense.