In recent years, researchers have increasingly turned their focus toward utilizing machine learning and deep learning techniques for crime detection. In this paper, using the PRISMA approach we identified 31 papers published in various journals/conferences in the years 2006–2024. This paper focuses on a few detailed research papers, exploring how computational techniques anticipate criminal activity. Emphasizing the importance of data sharing, it highlights collaborative efforts for collective learning and improvement. Through meticulous analysis, the review identifies effective computational strategies for crime detection while pinpointing areas for enhancement. The research offers access to the datasets utilized by researchers to predict crime. It also scrutinizes prevalent methodologies employed in machine learning and deep learning algorithms for crime prediction, providing valuable insights into various trends and factors associated with criminal behavior. It examines the strengths and limitations of different approaches, shedding light on which methods yield the most reliable results in different contexts. Ultimately, it serves as a valuable resource for refining crime detection models and empowering law enforcement with more accurate intelligence for safer communities.