Forensic Sketch Reconnaissance Using Deep Learning

Abstract

Contemporary society is passing an increase in ordinary crime. To fight this, law enforcement agencies must accelerate the entire process and find a way to impeach culprits against justice. One similar system could be using facial recognition technology to identify and corroborate culprits. Conventionally, forensic artists use hand- drawn sketches to identify culprits and contemporizing this system requires relating culprits by comparing the sketches to law-enforcement databases. Taking this approach will pose a number of limitations to the current technology, as there are fairly many felonious artists available compared to the growing number of crimes. Our idea is to speed up the process for law enforcement departments by creating a standalone platform that can be used to directly sketch a suspect without backing from a forensic sketch artist and with no special training or cultural chops. Sketches can be created with drag-and- drop in operations with colorful facial rudiments, and synthetic face sketches drawn using deep literacy and pall structure can be automatically counterplotted to law-enforcement databases much briskly and more efficiently.