Amazon recently announced a slew of improvements to its facial recognition software, dubbed Rekognition, along with a new feature; the ability to detect fear. This new feature is an expansion to the technology’s emotional gamut, which currently also sports happy, sad, angry, calm, surprised, disgusted, and confused. Adding fear to the mix is a logical progression, as ominous as it may otherwise seem. With the new emotion and improvements to previous ones, Rekognition will now have a reasonable handle on how its subjects are feeling when it deciphers their photos or live video feeds of them, much the same way a human observer might.
While Amazon was light on the specifics, it did mention that it uses “facial landmarks” in its detection process. Paired with the technology’s detection of things like a subject’s probable gender, possible age range, and “face pose”, it’s easy to see how a model could simply pick up all the features of a recognized face, compare a given configuration to what it knows of how a certain emotion may affect facial expression, and work backwards from there via validation to see if the emotion it guessed jives with a subject’s facial expression.
Amazon is focusing on improving Rekognition at the moment, because of its massive potential. Already, the company is seeing some return on its investment as law enforcement agencies jump on the bandwagon. Since the tech’s inception, Amazon has made no secret of its intentions to license Rekognition out to government agencies and the like.
Given the current sociopolitical climate of the United States, Amazon’s home country, one would be hard-pressed to argue that Rekognition won’t be used for mass surveillance. Since Amazon does business internationally, of course, it’s also quite possible that Rekognition will end up in the hands of other governments, some of whom are already engaged in mass surveillance.
That’s not to say that Rekognition is all bad news. The technology has incredible potential for a number of applications, such as VR avatars emoting, enhanced video chatting, and other human interface tasks. Likewise, Rekognition could end up doing genuinely great things, such as locating missing people, automating the tedious task of ticket counting for events, map facial genes, and more. The tech has made leaps and bounds in regards to its abilities and use cases, a far cry from the start of 2018, when it was so bad at its job that shareholders were calling for it to be shuttered.
At the bottom line, Rekognition’s expanded abilities make it a force to be reckoned with in the facial recognition world. Amazon’s facial recognition technology seems poised to usher in a new era of brand new use cases for facial recognition technology, while enhancing current ones. This, of course, is all speculation. The tech world can be wild, and even in the top echelons, anything can, and often does happen. With its potentially Orwellian aspects, after all, Rekognition is an easy target for privacy advocates and other sorts of activists, as well as regulating bodies the world over.