Like many feature-based forensic comparison methods, forensic facial comparison is based on subjective, perceptual observations made by trained expert examiners. These qualitative observations can be hard to validate and can suffer from poor repeatability between different expert examiners.

Our recent consultancy work has been investigating the use of state-of-the-art automated facial recognition technology and statistical modelling to produce a numerical, probabilistic output that could be used as forensic evidence (a likelihood ratio).

This innovative project combines the speed and accuracy of biometric technology with the transparency and balance of forensic expert examination, paving the way for future solutions in forensic biometrics.