Computer Vision in Smart City Application: A Mapping Review

Research interest in the application of Computer Vision (CV) in the context of smart cities has grown significantly in recent years. However, finding relevant studies in databases can be challenging due to the use of non-explicit or varying terms. To address this issue, this research aims to conduct a literature review by collecting papers that discuss the application of CV for smart city needs. The method used is a mapping review, with the following stages: 1) Establishing research questions; 2) Selecting relevant reference sources based on relevant databases; 3) Identifying appropriate keywords; 4) Conducting screening based on criteria; and 5) Extracting and synthesizing relevant data. For this research, we selected a total of 35 papers as the primary review. Our findings show that research trends and interests vary widely, with the most prevalent being in smart mobility components, which on average come from the IEEE Xplore Digital Library database. YOLO and CNN algorithms are also widely used to solve problems. Furthermore, we conclude that there are many benefits to be gained from the application of this technology to support the sustainable development of smart cities, including ensuring the quality of life of their inhabitants.

I. Introduction

In the era of ongoing digital transformation, the use of computer technology in the smart city context has become increasingly profound and relevant. According to [1] if viewed from the perspective of the use of Information and Communication Technology (ICT), the smart city concept seeks to increase efficiency and productivity, share data with the public, and improve government services and citizen satisfaction. smart city uses a combination of low-power sensor devices, cameras, and Artificial Intelligence (AI) algorithms to monitor city management and conditions, while [2] is also able to collect and analyze large amounts of data. CV is one of the technological trends in smart city. Advances in CV in terms of recognition and tracking have resulted in efficient image capture, image processing, and object recognition capabilities for vision applications.