AI-Powered Crime Scene Analysis

The project is about automatic analysis of indoor crime scenes. We will study AI technologies for environment mapping, segmentation and classification of objects and traces found at such scenarios worthy of immediate investigation, eg to avoid contamination of the scene or warn for hazardous situations.

Summary

Expected results and effects

Crime scene investigation is normally done by forensics experts upon arrival. The present project will automatize these tasks, allowing the team to directly concentrate on the analysis of important cues, thus saving precious time during the first moments after a crime. Outputs will also remain as an uncontaminated model of the scene, allowing post-analysis, if necessary, during any step of the investigation.

Planned approach and implementation

To achieve our aims, we will explore vision technologies like visible, IR, thermal, and non-vision depth sensors like LIDAR. To ensure that the scene is contaminated to the least possible extent, we will investigate the use of nanodrones. This is a challenge, since existing drones equipped with those sensors are usually big and unsuitable, for example, for small flats. To counteract potential difficulties in such innovative task, we will also investigate the use of smartphones or bodycams worn by first-responders.