Prediction Models for Car Theft Detection Using CCTV Cameras and Machine Learning: A Systematic Review of the Literature

Méndez Cam J.R.; Santos López F.M.; Rosales Urbano V.G.; Santos de la Cruz E.G. · Lecture Notes in Networks and Systems (2023) · DOI: 10.1007/978-3-031-30592-4_14

Resumen

Car theft is a constant problem in parking lots and places where cars are left unattended. Car theft detection is a time-consuming task due to the human resources that are required. Therefore, the task of checking closed circuit television (CCTV) cameras can be automated using machine learning techniques. The implementation of such a system would mean an optimization of the current technology. Even if a CCTV camera is installed, it requires human labor to supervise the area, which is a repetitive and time-consuming task. A machine learning algorithm could simplify the task and attend to many cameras without decreasing the attention on each one. In this context, a systematic review of the literature on machine learning was conducted based on four research questions using the PRISMA methodology. The research method may help to find the current methods used in similar applications and possible ways to implement the proposed automatic solution. This scientific study retrieved 384 articles from Web of Science, Scopus, and IEEE databases. The number of studies used to answer the research questions was 58. Finally, analyzing the most frequent models and metrics, Convolutional Neural Networks and Accuracy were the most referenced, with 30 and 42 mentions, respectively.

Instituciones: Universidad Nacional Mayor de San Marcos; Pontificia Universidad Católica del Perú