Towards automated eyewitness descriptions: describing the face, body and clothing for recognition

ABSTRACT

A fusion approach to person recognition is presented here outlining the automated recognition of targets from human descriptions of face, body, and clothing. Three novel results are highlighted. First, the present work stresses the value of comparative descriptions (“he is taller than … ”) over categorical descriptions (“he is tall”). Second, it stresses the primacy of the face over body and clothing cues for recognition. Third, the present work unequivocally demonstrates the benefit gained through the combination of cues: recognition from face, body, and clothing taken together far outstrips recognition from any of the cues in isolation. Moreover, recognition from body and clothing taken together nearly equals the recognition possible from the face alone. These results are discussed with reference to the intelligent fusion of information within police investigations. However, they also signal a potential new era in which automated descriptions could be provided without the need for human witnesses at all.

KEYWORDS:

• Biometrics
• eyewitness statements
• body description
• face description
• clothing description
• fusion

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Sarah V. Stevenage http://orcid.org/0000-0003-4155-2939

Notes

1 The term “labels” is used throughout the current manuscript. This may be interpreted as being analogous to the term “attribute” used in the computer vision and biometrics literature cited here.