Computer Vision Based Smart CCTV Solution

Abstract

Abstract With the assessment of the latest technology safeguarding humanity for their safety and security round the clock is an immense challenge. Deep learning algorithms are being developed to provide greater performance, accurate classification, and the extraction of useful information. Numerous deep learning architecture techniques have been put forth in recent years. Massive security camera deployments have been made, together with additional sensors, to monitor important infrastructure, including Oil & Gas industries, military bases, airports, power plants, banks, etc. As human resources are expensive and have a finite skill set, manual monitoring by human operators is an ineffective solution in modern world. The goal of AI technology used in CCTV system is to automatically monitor infrastructure or the environment with little to no human involvement. These monitoring activities involve automatically locating and tracking objects (such as people or vehicles), followed by additional analysis and action. Techniques for artificial intelligence (machine learning), signal processing, and image processing are key components in the development of such intelligent systems. The most popular modality (device) for a surveillance system is a visible camera like a CCTV. It has been used for a very long time to keep an eye on people and events. To automatically evaluate data (image or video) from surveillance cameras, extensive research has been done. Numerous specialized review papers on background-foreground segmentation, objects detection and classification, tracking, and behavioral analysis have covered a large portion of these investigations. Other than visible cameras, many sensor modalities have been investigated for surveillance systems, including thermal and infrared cameras, radar, and audio sensor. This paper will present the case study of a comprehensive overview of computer vision base CCTV surveillance system and to review the current approaches to each of its processing steps.