Researchers developing facial recognition wearable for emergency service workers
Researchers in the Department of Electrical and Computer Engineering at the University of New Mexico led by Professor Manel Martínez-Ramón are working on facial recognition-equipped wearable devices to help firefighters and emergency service workers with navigation, communication, and threat assessment, the university announced.
The projects are funded by a National Science Foundation grant called Next Generation Connected and Smart Cyber Fire Fighter System.
Semi-supervised facial expression recognition using reduced spatial features and Deep Belief Networks is the most recent paper which proposes a biometric facial recognition system to help firefighters identify facial expressions and emotions with 98 percent accuracy.
The team says this is the first solution to leverage semi-supervised learning where a network of computational nodes is trained to recognize faces by feeding it both labeled and unlabeled images, teaching it which facial expressions and emotions match.
According to Martínez-Ramón, the algorithm was tested using CARC (Center for Advanced Research Computing) resources. The ultimate goal is to integrate an accurate biometric facial recognition algorithm with a wearable device to help firefighters in emergencies. When faced with poor visibility, heat or imminent danger, firefighters can use the wearable to accurately identify victims. The goal is to create a supercomputing solution that can be used across the country, and that will use microphones, cameras, body sensors, and ambient sensors to monitor oxygen levels, potential hazards and the presence of victims, among others.