US Army is working on a New Thermal Facial Recognition Technology
How efficient are Thermal Facial Recognition Scanners
Facial recognition technology is now common in a growing number of places around the world from public CCTV cameras to biometric identification systems in airports already touching half of the global population on a regular basis. Some months ago, the startup FDNA created the DeepGestalt algorithm to identify genetic disorders from facial images. More recently, US Army published a pre-print paper documenting the development of an image database for training AI to perform facial recognition using thermal images.
In technical parlance, facial recognition is an artificial intelligence-based biometric technology that assesses a person’s facial features to identify or verify them from an image or video. The system then analyzes and compares the information with a data bank of other faces to detect a match. The facial recognition market surged owing to its non-invasive, contactless, quick, and accurate screening, especially during COVID-19.
When this facial recognition software is paired with the capabilities of a thermal scanner, the resulting thermal face recognition tool can be used in low and no light conditions. Typically these thermal scanners capture heat (radiation in the mid-wave infrared (MWIR) and long-wave infrared (LWIR) spectra) emitted by a living body. Depending on the brand and manufacturers, some of the thermal face recognition tools have an accuracy of more than 95%. It also addresses the limitation of the facial recognition system, which predominantly focuses on the visible spectrum. Using these scanners can enable the military in carrying out their covert night-time operations. Apart from that thermal face recognition cameras are used for surveillance in checkpoints and watchtowers and increasingly on body cams.
The US Army’s corporate research department, DEVCOM, released the DEVCOM Army Research Laboratory Visible-Thermal Face Dataset (ARL-VTF), which is the largest collection of paired visible and thermal face images to date. However, having only 600K total pics and 395 total subjects it’s actually relatively small compared to standard facial recognition databases. Although researchers at DEVCOM assert that insufficient data amount does not mean that the AI will be incapable of identifying faces effectively.
As per their paper, researchers have mentioned extensive benchmark results and analysis on thermal face landmark detection and thermal-to-visible face verification by evaluating state-of-the-art models on the ARL-VTF dataset.
“Analysis of the results indicates two challenging scenarios. First, the performance of the thermal landmark detection and thermal-to-visible face verification models were severely degraded on off-pose images. Secondly, the thermal-to-visible face verification models encountered an additional challenge when a subject was wearing glasses in one image but not the other.”
In case if this technology is further developed, it could result in better combat control on the battlefield. Though if unfortunately, the results do not improve, it could lead to the death of innocent people that are falsely recognized.
For more info on this, please read here.