US Army Research Lab Creates Low-Light AI Facial Recognition

Artificial Intelligence AI Brain AH

The US Army Research Lab has created a low-light AI facial recognition software system that makes it possible for the system to see and recognize people’s faces in the dark. The idea behind the technology is geared at military use and aims to make it possible for facial recognition to happen without the need for a light source in a particular situation where lighting may not be the best, and may have otherwise been needed to match facial imagery to already existing databases. With this technology available, night time operations where one objective would be to match the facial imagery to a database would be significantly easier, at least in regards to the facial recognition.

According to the ARL researchers the technology works through thermal-to-visible face synthesis so it can match those thermal facial images to the data base of visible facial images. While the technology is likely far from complete it was reportedly demoed recently with near real-time results using a FLIR Boson 320 thermal camera and a laptop that the algorithm of the AI software.

Applications for this kind of technology could be used for various operations including human-matching capabilities as well as auto-matching capabilities from software, both in real-time and after the fact, though real-time use has not yet been achieved in testing. In theory this could be used for much more than just military applications, and could be hugely beneficial to any industry where facial recognition might be a tool that is already being used. Law enforcement for example could also benefit from the ability to match thermal facial images to visible facial images when sufficient light sources aren’t available. The US is not the only region experimenting more with facial recognition software. Earlier this month it was reported that Chinese authorities were able to use facial recognition software to find and apprehend an individual amongst a crowd of over 60,000 people