Recent Implementation of Deep Learning Methods for Shoplifting Detection in Retail Environments Using Surveillance Videos

Abstract

Shoplifting is a problem that continues to face retail businesses and is associated with a serious loss of revenue and security issues for retailers. The recent developments in deep learning have created new possibilities to develop intelligent surveillance systems that detect suspicious behaviour automatically and with high precision. This paper systematically reviews recent literature between 2018-2025 on deep learning methods for shoplifting detection in retail environments. It discusses the applications of deep learning mechanisms in the detection of shoplifting in retail settings. In addition, the paper presents methods that utilise Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and hybrid deep neural networks to detect shoplifting. Further discussions were presented on privacy-preserving architectures, behaviour analysis, and anomaly detection using surveillance videos. To ensure continued implementation by researchers, the paper outlines research gaps such as low levels of real-time use, poor general applicability in a variety of retail environments, low compliance with privacy, and ethical issues that require further research. The findings reveal that deep learning approaches are important artificial and machine learning techniques for shoplifting surveillance and detection of suspicious activities in retail settings for prompt prevention of crimes and thefts by business owners. Also, the study will add to the contained improvements to the existing body of knowledge in the field of computer vision, deep learning, and cybersecurity.