Violence detection in video surveillance systems is a critical task that has garnered significant attention across research and industry domains. This literature review explores various approaches to violence detection in video surveillance systems, encompassing traditional methods using handcrafted features and advanced deep learning models. This review provides a comprehensive exploration of diverse methodologies employed in this field, ranging from fundamental approaches leveraging handcrafted features to cutting-edge deep learning models. The review covers motion-based detection methods, deep learning-based approaches, hybrid models, comparative analyses of methodologies, and real-world applications and datasets. By examining these diverse techniques, the study aims to highlight the strengths, weaknesses, and future directions in violence detection research. The review categorizes these methodologies into motion-based detection methods, deep learning-based approaches, hybrid models combining multiple techniques, comparative analyses of methodologies, and practical applications using real-world datasets.