VisPer

Person recognition is already being used successfully in various applications, such as biometric access control or real-time surveillance of public spaces. In forensic analysis, however, person recognition continues to pose a particular challenge. Relevant features are often only partially visible – due to obstruction or low image resolution, for example – and are also subject to changes over time, such as aging, hairstyle, body shape, or clothing style.

The VisPer project investigates how individuals can be recognized under these and other challenging conditions using modern computer vision methods. So-called soft biometric features – such as tattoos or scars – which remain stable over long periods of time, play a central role in this process. To effectively support forensic experts in their case work, person recognition must be flexibly adaptable to different forensic scenarios and deliver transparent, interpretable decisions.

For this reason, the system developed in the VisPer project is specifically designed for collaboration with forensic experts. Criteria for person matching can be adjusted on a case-by-case basis, while the decisions made are transparently justified and thus remain verifiable by humans.

In close collaboration with law enforcement agencies, VisPer is developing new and effective methods for person matching in forensic analyses. Possible application scenarios include, among others, the identification of missing persons and the recognition of victims or suspects.