Of all the features included in video doorbells, face recognition is perhaps the most underrated. There’s nothing quite like hearing your smart speakers yell “Mom’s at the door” before she even knocks. Unfortunately, people have a good reason for rejecting this feature – it doesn’t work.
Visit any smart brand’s support forum and you’ll see a ton of people complaining that their smart doorbell misidentifies every guest. Video doorbells think toddlers are grandpas or that every pizza delivery guy is your spouse. And there is not much you can do to solve this problem. Smart doorbells simply fail at face recognition.
Doorbell cameras don’t capture enough detail
The face recognition technology used by smart doorbells is quite modern as it uses some of the most advanced AI ever developed. But even with great AI, video doorbells don’t have the hardware to really support facial recognition – they just can’t capture enough detail.
Like most facial recognition systems, smart doorbells capture and analyze 2D images. These flat images contain a lot of useful information, such as: B. the width of your mouth, your skin tone and the distance between your eyes. But this data may not be unique to your face. In fact, this data may not be as accurate as video doorbells capture relatively low-resolution images of moving objects.
More advanced facial recognition systems, like the one in your iPhone, capture “3D” images using infrared TOF cameras. Here’s the gist; You shoot invisible lasers at your face and measure the time it takes for each laser to bounce back. The data captured by these TOF cameras contributes to a “depth map” that includes measurements like the length of your nose or the angle of your ears.
This 3D data is far more useful than what your video doorbell captures, for obvious reasons. But hardware isn’t the only problem here. On the whole, the advanced AI used by your video doorbell is actually pretty rudimentary.
Face recognition algorithms need to be trained
The facial recognition systems offered in smart doorbells are “self-learning”. They might ask you to identify a new person, but for the most part, they create and organize a face database with no user input. And that can cause some problems.
You see, self-learning facial recognition systems are always trying to improve their accuracy. That means collecting a lot of data; It’s difficult for your doorbell to identify someone if it’s only seen that person once. Every time your doorbell sees “Mom,” it adds to its collection of This Is What Mom Looks Like.
But when a plumber comes to your door and is mistakenly identified as “Mom,” the facial recognition system becomes less accurate. Her doorbell doesn’t know she made a mistake, and suddenly guests with a mustache could be “mommy.” This leads to a downward spiral – the loss of accuracy leads to more false positives, and “Mama” now comes in every size, shape, and skin color. In the eyes of your doorbell, everyone is “mom”.
It’s like a student learning a math equation wrong. They may rely on what they’ve learned, but until they bomb an exam, they won’t realize they screwed up. Students need someone to check that they are learning things correctly, and the same goes for the AI.
Unfortunately, in this situation, you are the teacher.
How to improve face recognition on your doorbell
Improving your doorbell’s facial recognition system is a chore. There’s no permanent fix here – you’ll have to actively keep up with the facial recognition system to fix its bugs.
The first thing you need to make sure is that your smart doorbell can clearly see guests’ faces. That might mean repositioning the doorbell, cleaning the lens regularly, or adding some lights on your front door.
Once you know your doorbell can see what you want it to see, you need to clean up its facial recognition database. This process is different for each doorbell, but in most cases you can find a list of faces in your smart doorbell’s companion app. (If you have a Nest doorbell, go to the Nest Aware settings in your Google Home app. I don’t know why Google hides these things.)
Delete any misidentified faces your doorbell has picked up, and be sure to tell your doorbell the names of any unidentified people you expect to return to your home. If you maintain this database regularly, your doorbell should get much better at identifying guests.
Here’s the bad news; Even if you try to improve the accuracy of your doorbell’s facial recognition system, it will never be perfect. Actually, it can always suck. Face recognition technology is still fairly rudimentary, and smart doorbells use very basic hardware to “see” people.
If you hate the idea of curating a facial recognition database, you might want to just disable the feature. Replacing your doorbell isn’t worth the money as all brands’ facial recognition systems suffer from the same issues.
This article was previously published on Source link