A researcher at the University of Malta recently designed a system that automatically retrieves the identities of suspects portrayed in sketches described by witnesses to a crime, through the comparison of such sketches with traditional photographs.

[attach id=687926 size="medium" align="right"]Christian Galea[/attach]

The research project, entitled Face Photo-Sketch Recognition using Deeply-Learned and Engineered Features, was undertaken by Christian Galea as part of a PhD in ICT degree recently conferred on him by the University of Malta.

The aim was to design a system that could perform facial recognition using different inputs, and despite the inaccuracies of facial sketches and eyewitness descriptions. The system implemented several novel methods, ranging from the use of traditional engineered features to advanced machine learning techniques. A database of facial sketches was also created using the same software used by law enforcement agencies worldwide, including the Malta Police Force.

The designed system exceeded state-of-the-art performance of a variety of sketches, including forensic sketches.

The research has been published in leading peer-reviewed international publications.

Dr Galea is currently a post-doctoral researcher at the Institut National de Recherche en Informatique et en Automatique (INRIA) in Rennes, France, and in February he will start working at the University of Malta as part of the Deep-FIR project aimed at improving the quality of facial images captured by CCTV systems.

Dr Galea carried out the research under the supervision of Dr Reuben Farrugia at the Department of Communications and Computer Engineering of the University of Malta’s Faculty of ICT, and in collaboration with the Malta Police Force.

NVIDIA Corporation donated a Titan X Pascal graphics processing unit for use in the research. Dr Galea’s PhD degree was supported by a scholarship awarded under the Malta Government Scholarship Scheme (MGSS).