Facial Recognition – The Process In 3 Steps

Face Recognition Process by Corsight

Three steps to facial recognition? It sounds simple and with the right technology it probably is simple.

But if one of these steps leads to excessive errors – either false positives or false negatives – it could be very troublesome for your business. False positives are particularly problematic.

What are these three steps? How can you make sure they work towards a high level of confidence in the results?

Step 1 – Detection

This part of the process filters out the faces from all the things within range of the camera. Of course the type of camera(s) and the filtering technology or algorithms are vital to this effort. The vendors of systems you are considering should be able to demonstrate their systems’ capability – your responsibility is to put them to a rigorous test when you have a demonstration. Make sure you are satisfied the system will work under the conditions – lighting, movement, attire (eg medical masks) – it will produce reliable results in those conditions.

Step 2 – Image Capture

The facial recognition system will take the characteristics of the detected face and convert them into a digital format. Clearly this technical process will be difficult to test to your satisfaction. To some extent you will have rely on the results you see.

Step 3 – Matching

This step confirms that the digital information representing the captured image is a match to the data obtained from information already stored … once confirmed an alert or notification is issued. This is the step that could land you with a false outcome. A false positive will incorrectly identify the person before the camera, a false negative will fail to match a person that should have been matched.

In the hospitality context the stored images are most likely the images of members / guests / customers held on Membership or CRM systems. Whatever facial recognition system you install these images will have a huge impact on the accuracy of the matching process. Essentially, this is one of those things that falls under the old technology saying – garbage in, garbage out!

There are definitely actions you can take to make sure you do not fall into this trap – the actions will depend on the type of CRM or Membership system you are using.

Before Acquiring a Facial Recognition System …

Checking on these 3 steps after you’ve purchased your facial recognition system is … Well, the horse has bolted and you’ll be stuck trying to retrospectively improve on the results. By the time you realise the need, you will likely have experienced some fairly damaging instances of incorrect identifications.

So, before making your decision to purchase:

  • Arrange for a live demonstration of each system at your venue.
  • For the demonstration make sure:
    • the lighting and environmental conditions are as close as possible to those the system will need to operate under,
    • you have a list of items/questions you want clarified or explained and ensure they are covered to your satisfaction,
    • have your database, CRM, membership system checked to determine the if the images will help achieve a high level of confidence in the matching process.
  • conduct any other due digilence you require of suppliers / vendors. Remember, you will undoubtedly need support from the supplier and having a good working relationship is very valuable.

We would love to put our Facial Recognition system “through the hoops” at your venue. We are confident that we have the best system … and we can help you with working out how to make sure you can have confidence in its connection with your membership system.

* feature image courtesy of Corsight