Real-time Activity Detection & Recognition in Video

Abstract

Real-time video recognition and identification is the ability to analyze video automatically to identify
and evaluate temporal events that are not dependent on a single image. It is the process by which a video
is processed, data gathered and data analyzed to obtain domain-specific information. Object Detection
is the method of identifying instances of real-world objects. It takes into consideration to recognize,
localize, and detect multiple objects within an image, in video real-time. This paper is about a security
monitoring system that uses the identification activity of the human body to build improved surveillance
and also detect suspicious objects. Nevertheless, no unusual circumstances are detected by people, rather
computers that analyze the recorded pictures and identify suspicious behavior or incidents. In this paper,
we have described a system to distinguish and perceive suspicious activity. We are using the OpenPose
Python API for our conceptual development