Establish an AI Operating Model and get out of prototype and into production
March 1, 2019

How facial recognition and the right AI provider can improve your operations

Table of Contents:

Facial recognition’s presence in our everyday lives has grown immensely over the last few years. From tagging friends on our social media pages to unlocking our mobile devices, few iterations of visual recognition are more popular or versatile.

Still, with the use-cases we see in the media (like Popsugar's Twinning app,) this computer vision technology might seem very out of place for the practical needs of operations managers. On the contrary, however, they are among the stakeholders who stand to gain the most from investing in this AI.

Below is a simple guide on facial recognition and how it can be used to help improve your operations. Before we get into it though, let’s first take a look at the step that must take place before facial recognition can get going: face detection.


What is face detection?

Face detection is a computer vision technique where image recognition detects the presence of faces within an image, then locates them by placing a box around each one. It can be used to build computer vision “models” or data processing blocks that take inputs, like images, and return certain "concepts” or tags.

To put it simply, face detection tells us WHERE the faces in the image are.


Screen Shot 2019-02-28 at 4.54.52 PM


Operations use cases for Face Detection

When combined with your security cameras, face detection can help you to quickly and accurately measure how many people visit your store daily. Comparing this count to the number of purchases actually made could, in turn, help you to identify whether there are opportunities for improvement. It could also help you to better determine your peak visiting hours by counting the number of customers in the store at any given time. You can then use that information when scheduling your employees, to ensure you meet your customer and efficiency needs.


So, what is facial recognition?

Facial recognition (also called “face recognition”) is a computer vision technique that can be used to identify people. Much like the human brain, a trained custom face recognition model can see a person’s face and learn to recognize them. It can then use that knowledge when it sees a new image of a face to quickly determine whether this is a person it knows or a stranger.

For this technology to work, you first need to gather several images of each person you want the AI to recognize. Like most humans, AI has to see a person a few times (and at a few different angles) before it can learn who they are.

Once a facial recognition model is trained, deciding whether the face seen in an uploaded image is one it’s seen before is a simple, “one-step” process. That is, this technology will know immediately whether or not this is the face of a stranger, without needing to compare it with all the other faces it learned.

To put it simply, facial recognition tells us WHO is in an image.



Operations Use Cases for Facial Recognition

When applied to customer service, facial recognition can help you by learning to recognize your most frequent or loyal patrons, so your employees can be quickly alerted when they enter your store. This will allow you to create the personalized experience customers are now expecting.

It can also be used to help you manage and oversee your team, ensuring the right person is in the right place, at the right time. For example, depending on your business’ policies, only certain employees may be authorized to enter particular parts of your store or warehouse. While many businesses currently rely on magnetic-stripe keycards to secure these areas, these cards can be lost, stolen, or damaged, potentially creating a security or safety hazard. With facial recognition, you can mitigate this issue without compromising security or speed of entry, by teaching your model the faces of who should and should not be allowed to enter. If an unauthorized person attempts to gain access, the relevant security personnel can then be quickly alerted.


Build vs. Buy: Why you should work with a computer vision provider

Unless you are an AI company, building and training any AI in-house, including face recognition, is very costly and impractical. You will need to not only pull together the right team of experts to build this technology but also to maintain and improve it. You’ll also be competing with tech giants and AI start-ups for talent that is still relatively scarce and so, have to redirect valuable resources away from the staff your core business needs.

Working with the right image recognition company removes these problems. They’ve already built the required infrastructure for your AI platform with their existing team of experts. All you will need to do is provide the data (i.e., images of faces) and concepts (like names or customer type) you want your custom facial recognition model to learn. The best visual recognition partners will work with and guide you every step of the way, from implementing the platform to maintaining and upgrading it without your intervention.

To put it simply, for most businesses, using the appropriate third-party AI provider for your facial recognition needs is the best option. They will be your team of AI experts, so you can focus on what really keeps your operations running smoothly.

Need help deciding? Download our checklist!

Ebook - Leveraging AI to Drive the E-commerce Experience


Are there alternatives to facial recognition?

While facial recognition is not the only computer vision technology that can be used to identify persons, it is by far the one that is most worth the investment. It might be more expensive to put in place than other techniques, like face comparison, but it is also a lot faster, accurate, and flexible than those techniques.

3806778-3x4-700x933For instance, while face comparison does not require “training,” that means it doesn’t truly learn what the face actually looks like. Instead, this image recognition technique only analyzes images to see whether they are visually similar. For example, while you may have the photos from your employee ID cards in a database, even if one of your employees is just not looking in the same direction, a face comparison model may not be able to identify them, as the new image and the one in your database may not be visually similar.

Since facial recognition requires your training data to include several pictures of each person, such a model will be able to recognize the person, even when they are looking left instead of right or wearing blue instead of red. Further, while face comparison can sometimes be cheaper than facial recognition, this is not always so, a matter of importance considering how much less accurate and flexible it is compared to face recognition.

 While many computer vision companies may call their product offering “facial recognition,” be sure to ask what their product can actually. Many companies, for instance, use “facial recognition” to describe their entire suite of “face products,” including face detection. Others may also use “facial recognition” to describe “face comparison,” which you now know isn’t the same thing.


Facial recognition might sound too high-tech or excessive for your needs, but that’s only if you try to build your own facial recognition technology from scratch. By working with the right visual recognition provider, you will be able to use one of the most cutting-edge AI technologies available today in a way that is practical and affordable for your needs and your business.


 Ebook - Leveraging AI to Drive the E-commerce Experience