Applications of Artificial Intelligence in On-Site Crime Scene Analysis: a systematic review of Automated Forensic Techniques

The crime scene represents a pivotal moment in the investigative process—complex, dynamic, and unrepeatable. It is the first point of contact between the criminal act and the judicial system, and the foundation upon which evidence collection, event reconstruction, and courtroom narratives are built. The quality and epistemic robustness of the sources of evidence gathered at this stage critically influence the entire judicial trajectory. Yet, despite its recognized centrality, crime scene analysis remains fragile: lacking international procedural standards, subject to contextual variability, and heavily reliant on the subjective interpretation of forensic professionals. These vulnerabilities compromise replicability and reliability, contributing to judicial errors. Compounding these issues are systemic challenges—high costs, personnel shortages, inter-laboratory heterogeneity, and delays incompatible with investigative urgency. In this context, artificial intelligence emerges as a transformative ally. Techniques such as machine learning, deep learning, and computer vision offer the potential to automate complex tasks, standardize procedures, and enhance analytical accuracy. Far from replacing human expertise, these tools aim to augment it, fostering a more objective, scalable, and scientifically grounded investigative process. This systematic review critically examines the current landscape of AI applications directly at the crime scene, assessing their effectiveness, maturity, and integration potential, while identifying methodological gaps and future directions.