Abstract

Crime scene investigation is a significant part of forensic science; the proper search and analysis of evidence could alter the direction of the case. Traditional approaches involve working on a manual document, which is prone to both inaccuracy and evidence falsification, as well as waste of time. AR is a recent tool that involves superposition of digital data on real scenes to improve visualization and analysis. This review includes the latest articles on AR in the forensics field, which comprise 30 articles (2019-2025) that mention methods of object recognition such as YOLO, data integration to educate a model, Siamese networks to match an object, sequence guidance, and profile suspects with AI. We feature the techniques of AR overlays, AI fusion and advanced features. Such comparisons are made on data sets, measurements and benchmarks, and constraints to these comparisons are also discussed on a critical note. These problems are discussed: real-time processing, privacy, and such prospective ideas as targeted alerts and evidence links. This serves to direct researchers and investigators to apply AR to make a more valid analysis of crime.