1. Home/
  2. Blog/
  3. Celebrity face recognition: make streaming just a little bit better

Celebrity face recognition: make streaming just a little bit better

Celebrity face recognition: make streaming just a little bit better

Details, people, details! When it comes to delivering video content to the users, it’s the little things that can make your platform truly stand out.

Having all of the content in the world does not cut it anymore. Remember Blockbuster? There you go.

Today we will give you a little hint on how you can make it easier for users to enjoy your content and find something they like — with celebrity face recognition by Cognitive Mill™. Let us show you how.

The deal with celebrity face recognition

I’m not that great with remembering faces. Not good as an eyewitness for sure. And when it comes to recognizing celebrities, I can sit in silence in front of the TV for minutes trying to remember the character's name and in which show I have seen them before.

With all of us indulging in more video content than ever, I can see this being quite relatable. It becomes increasingly hard to keep track of everybody cast in all of our favorite shows.

The struggle with celebrity face recognition is real.

Famous sports players and politicians constantly appearing on TV only make the problem worse.

From the platform point of view, a ton of people appearing in tons of video content poses another kind of problem: it becomes harder and harder to effectively manage all of the assets that a TV channel, a broadcasting company, or an OTT platform has.

But there is something that we can do about it.

And all of those famous personalities can help us with both problems.

Applying technology to celebrity face detection

Artificial Intelligence is the thing that can spare us from cracking our heads open trying to remember who exactly we see on the screen.

In addition, technology can also guide us towards more content involving a particular celebrity. The platform gets more views, the viewers get more content — it’s a win-win from where I see it.

And when it comes to the platform, celebrities’ info is a convenient way to store and manage the video content more efficiently. Tag all of the content by a famous person involved with it and you have a nice little system that you can use to categorize, navigate, and access the content with ease.

Celebrity identifier makes it easy to organize and access content.

Now, onto the real problem here: how do we get technology to identify celebrity faces amongst countless extras over thousands of video minutes?

Face recognition APIs are the things that first come to mind, right?

The technology widely used for authentication purposes was adapted to recognize celebrities' faces appearing in video content by several technological giants.

Their respective solutions are simple face recognition algorithms paired with an indexed gallery of celebrity faces. Here’s how that process goes:

Would like to try celebrity face recognition?

  • The algorithm is fed a video clip for analysis;
  • It decomposes the clip to apply face recognition;
  • The solution compares the found faces against the indexed gallery to find matches;
  • The tool tags the faces it has recognized.

And we can already see several problems associated with that approach.

AI gonna AI

The most straightforward approach is not always the most effective one. The issue with Artificial Intelligence being used for celebrity recognition is its accuracy.

Despite having a library of faces to consult, the majority of AI-based algorithms still require a human operator to go through some of the marked content to ensure the correct analysis.

Simple AI face recognition still requires help from humans.

On top of that, is the algorithm we are talking about, which is not a complete solution applicable to media & entertainment use cases. It can recognize celebrities, but it can’t identify main and side characters, nor can it pull more content associated with a particular celebrity.

Talking about character identification — the cast of some movies and TV shows can be enormous having a ton of famous names. While there are only about 10 characters that drive the plot, a regular AI celebrity recognition will blindly compare all of the faces it has found to the whole gallery of indexed celebrities. Which is — you got it — more time for analysis.

So, what do we do then?

Celebrity identifier powered by cognitive computing

The answer is as straightforward as it gets: if a robot doesn’t do the job, you make a smarter robot.

And that’s what we did — we took celebrity identification a step further by utilizing complex cognitive computing technology, which is capable of working with insufficient datasets and human-like contextual understanding.

Yes, it’s Jennifer Lawrence from The Hunger Games, you can continue watching the movie.

Thanks to that, we were able to deliver an API that is perfectly suited for the media & entertainment industry.

Here are a couple of features that benefit said industry the most:

  • Accuracy. With the help of complex face recognition algorithms the solution is able to perform celebrity face recognition in video content with no help from human operators;
  • Contextual understanding. Apart from celebrity face detection, the tool also recognizes them as main and secondary characters based on their performance in the movie. This way, the viewer gets the info only on cast members important to a particular part of the story;
  • API integration. Cognitive Mill™ can be integrated with celebrity databases so that your platform could provide more information about the cast members to the viewers along with more content involving said cast. On top of that, you can use it to categorize the content by a certain celebrity, making it easier to organize and manage;
  • Cast previews. The solution can pull the photos of actors from the analyzed clips, so you can provide the cast information with representative frames;
  • Speed. Automated celebrity face recognition can process a movie or a sporting event in seconds. This way, you can save a lot of time on manual editing;
  • Scalability. The cloud platform can be scaled to process hundreds of minutes of video content per day.

In addition, Cognitive Mill™ is developed as part of the CognitiveMill™ cloud platform, which means close integration with other products. So, having recognized the celebrity faces in a particular video clip, you also create:

Explore the benefits of Cognitive Computing!

Automation makes streaming movies, TV shows, and sports easier for everybody. All of the aforementioned features make the process of producing video content and delivering it to the viewers the most effective.

Benefit from other CognitiveMill™ products: make celebrity-centered trailers or highlight compilations.

On the other side of things, the viewers’ experience with the platform gets facelifted. Now, they can just hit the pause button during a movie and see the cast members highlighted on the screen along with some info on them.

Upon visiting the preview page of a movie or a TV show, the viewers can look at the celebrities cast in a particular piece of content. And the celebrities have their images pulled from the very content the audience is previewing, so it is easier to recognize them.

Along with that, the viewers can access the best moments compilations with their favorite celebrities — that’s another way to boost engagement with the platform.

It’s not only about the scripted media — Cognitive Mill™ is smart enough to work with the content of different genres. So, the audience gets help recognizing not only actors but other public personas, like famous politicians or sports players.

Optimizing production and increasing engagement rates with celebrity face recognition

If we are talking about having an edge over the competitors in media & entertainment, having the content is going to win you the audience in the long run.

Several companies have tried and failed at keeping their viewership numbers offering good content but delivering a lackluster viewing experience.

The superb viewing experience is all about adding little things that make finding and consuming content as seamless as possible. The audience wants to spend more time watching and less time searching — makes sense, right?

It is important to make the content more accessible not only to the viewers but also to the platform that manages it. As the library continues to expand, it becomes increasingly hard to organize and manage.

And that is where celebrity face detection comes into play. With the help of that, the viewers can get timely information of a movie cast member just by hitting the button. They also can go ahead and view more content with those celebrities involved, which only adds onto the navigation flexibility.

An automated celebrity face identificator Cognitive Mill™ helps platforms cut costs on human labor while retaining the same level of accuracy. It integrates neatly into your production process to not slow it down, and also provides access to additional features like highlight generation or nudity filtering.

If you are looking to boost your production effectiveness, reach us at support@cognitivemill.com or fill in the contact form below. We will be happy to answer your questions and set up a product demo to show the whole process.

Related Insights

Read all