fbpx
Facial Recognition as a Service

Facial Recognition as a Service

I was recently asked to question: how could we implement Facial recognition into our systems?

As you can imagine, building a Facial recognition engine is quite a complex task. Where do you start? 

First, you will need some form of image recognition system that can locate human faces within an image. 

This would be an AI system that can take in an image and determine the difference between a human face and any other object. This would require a large amount of R&D and once you have developed the system you will need to train it to ensure the accuracy of it finding a human face. 

Now once this has been established you with the need to build the model to determine the difference between one face and another. We have all seen a photo that looks like this:

Placing data points on a face and determining the distance between the data points will set you on the path to building an AI engine that can recognise one face from another. But do you really need to go down this path?

Facial Recognition as a Service

A major advantage with how far we have come with technology is that we do not need to reinvent the wheel. With new business models, you no longer need to invest considerable amounts of money to develop this type of technology. 

Microsoft Face API

Microsoft Face API is a SASS project that allows you to take advantage of an existing facial recognition engine. It allows you to: 

  • Detect and compare human faces in images
  • Organise images into groups based on similarities
  • Identify previously tagged people in images
  • Run locally on-premises or in the cloud

It is such a great piece of technology that will give you advanced facial recognition software without having to invest in development.

I took the opportunity to test the technology directly from the Microsoft Website and here are the results.

As you can see, I have uploaded my LinkedIn Profile picture and an image of me running, let’s see what happens:

As you can see, the application returned and 84.349% chance that the two photos were of the same person. I decided to run another test. The next one is of me coming out of the ocean:

As you can see, the application returned and 87.308% chance that the two photos were of the same person. Given that they are two very different images I think this is very impressive.

To learn more about this service please visit https://azure.microsoft.com/en-gb/services/cognitive-services/face/.

With the ability to leverage facial recognition as a service what business processes could you improve by utilising this technology?

Leave a Reply