FaceMatch: a visual search system for family reunification during disasters

Authors

• E Borovikov
• M Bonifant
• G Lingappa
• M Gill
• G Thoma

Abstract

The objective of our system (FaceMatch) is to provide a fast and robust visual search capability for those who are looking for missing people in the aftermath of calamities, natural or otherwise. We propose a Single-Image-per-Person (SIP) approach to face image matching using an ensemble of weighted face image descriptors, taking advantage of their matching strengths. FaceMatch utilizes many visual features including color, texture and shape to provide robust face detection, matching and image near-duplicate detection capabilities. To test our system, we considered several well-known benchmark face image collections. Our results are comparable to, and in some cases, better than those of leading open source and commercial systems, especially for low quality images, which is usually the case in disaster scenarios. FaceMatch is integrated in a real-world family reunification system called Lost Person Finder (LPF). It has been used during several hospital drills and actual disaster events, providing an efficient visual search modality; extending the text-based search facility; and benefiting hospitals, national/international organizations, and other entities closely involved in handling such events.

Scientific Focus Area: Research Support Services

This page was last updated on Friday, March 26, 2021