Aging is a very interesting phenomenon. I don’t mean medically, but from how we look at the process. When we are young (say under 10) we know our ages down to the month. In our twenties too, we can confidently state our age, albeit after a quick bit of mental arithmetics. Enter the 30s and you need to start counting from your birth year to come up with an age. Well, machine learning could one day make all the fumbling redundant. Towards the same, the Amazon Rekognition service can now guess ages.
Although, if you are hoping for something that could guess your age within the hour, prepare for some disappointment. Like everything that is powered by neural networks and machine learning, Amazon Rekognition’s age guessing capabilities are still a work in progress.
As per a post on the AWS blog on the topic,
Amazon Rekognition returns an array of attributes for each face that it locates in an image. Today we are adding a new attribute, an estimated age range. This value is expressed in years, and is returned as a pair of integers.
Rekognition is actually a pretty useful AI-powered service. Capable of doing a whole lot of stuff like detecting objects, scenes, and faces in images, the service can also leverage the power of AI to compare between faces. The service is powered by Deep Learning and analyzes billions of images everyday in order to evolve and become better.
The ability to guesstimate the age of the subject in the picture is the latest ability picked up by the AI.
So for the following picture (which belongs to Jeff Barr):
The service returned the following results:
And yes, Barr in fact was two years old at the time. Along with age range, the AI also attempts to guess the gender and whether the subject is smiling or happy. While it guessed everything fairly accurately for the image above, things got a little more complicated when other images were inserted, with the age range expanding to over 17 years!
However, as I earlier said, the ability is a work in progress and will likely get better with time as it churns through more pictures. Such a feature, if and when it starts to work correctly, could have a wide range of applications in different scenarios.
You can use this new attribute to power public safety applications, collect demographics, or to assemble a set of photos that span a desired time frame.
Along with Amazon, others also offer the ability to have their AI guess your age — prominently Microsoft. However, its AI isn’t particularly better than Rekognition and both of them would appear to be at somewhat same stages in their learning curve.
Introduced last year, Amazon Rekognition is a fully-managed service brought to you by Amazon. The service is capable of recognizing a wide variety of things by leveraging Deep learning and a set of API functions. The company offers the API to developers as part of the AWS and folks can make use of the AWS Free Tier tier, that lets you analyze up to 5,000 images per month, for free.
Meanwhile, you can check out the new age-guessing feature out right here. Do share the results!
A bibliophile and a business enthusiast.