Law enforcement officers, military personnel and forensic analysts are increasingly reliant on imaging systems in order to perform in a hostile environment. From surveillance systems to computer forensics, intelligence personnel require a robust method to quickly and efficiently locate objects of interest in images and videos. Most current approaches require a full-time operator to monitor a surveillance video or to sift a hard drive for images that could implicate a suspect.

In this paper, we demonstrate the effectiveness of automated image analysis tools to spot weapons in videos through a standalone application and user-interactive analysis. By training multiple appearance-based classifiers on a large corpus of labeled data, and by combining classifiers through machine learning techniques, we can create an overall classifier to detect individuals holding an AK-47. By automatically locating people carrying weapons in video, operators can be tipped to suspicious events. Current results indicate the value and importance of automated threat detection for forensic analysis in support of the law enforcement and intelligence communities.

Menu

Justin Jones