The Technical University of Munich (TUM) is developing an AI solution in cooperation with the Bavarian State Criminal Police Office (BLKA) that can intelligently analyze digital crime scenes in the future. The goal of the research project is 3D crime scene models to be used not only for visualization but also to specifically search for evidence and spatial relationships using Artificial Intelligence.

Already today, the BLKA creates detailed 3D reconstructions of crime scenes from hundreds or thousands of photos. The new AI technology is intended to extend these digital twins with interactive analysis functions. Investigators will be able to search for specific objects such as knives or clothing, or have all potentially relevant objects displayed automatically.

The researchers are developing AI methods for automatic object recognition, classification, and inventorying. In the long term, various traces - from objects to spatial information - are to be combined in a common virtual crime scene. This will allow connections between objects, positions, and possible sequences of events to be recognized and understood more quickly.

Another focus is on the more efficient processing of large quantities of images. A newly developed method automatically recognizes photos without any additional information content and thus reduces the computational effort in creating the 3D models. Furthermore, the research team is working on AI functions for analyzing lines of sight and spatial relationships, for example, to check which areas of a room were visible from certain locations.

The developed procedures are to be gradually integrated into the workflows of the Bavarian and Hessian State Criminal Police Offices. In the future, the technology could significantly simplify the use of digital crime scene twins and even support mobile crime scene capture via smartphone.