Automated Real-Time At-Scene Reporting System
Akeem Olowolayemo, Pearly Oh Bei Qing · 2018
This paper proposes a framework for real-time tracking of objects in real scene using computer vision technique implemented on a mobile application for news reporting. This automated reporting system is intended to complement news reporters in reporting real-time scenario especially in dangerous environment such as disaster situation, fire, explosion or war zones. The system comprises assembly of some functionalities and algorithms to successfully build and run function by converting scene images into paragraphing and voice output to describe the scenarios in real-time. A variety of functions such as object motion detection, scene recognition, emotional recognition and distance detection are presented so that paragraphing and voice generation can be more accurate and close in meaning to real-time human reporting situations. Since the proposed system can facilitate news reporting in real turbulent situations, the burden of news reporting can be reduced under such dangerous circumstances. This work briefly reviews the rudimentary concepts of image processing and computer vision that serve as the components of the proposed real-time automated reporting system and describes how these various compelling applications are coupled and work together. The work also outlines the choice of design for effectiveness and efficiency of such reporting systems.