Discover our knowledge. Read our blogs!

Learn more

We build all our solutions with WSO2 and we are proud that we are WSO2 Premier Certified Integration Partner and Value-Added Reseller.

Learn more

How visual recognition APIs are both useful and awesome

6 min read

Ever asked an IT guru what’s new in software? You’re probably flooded with stories about futuristic inventions that are too expensive to even consider buying. This can be tiring, as worldwide internet and brains that are attached to smartphones will not exactly boost your business (at least not until 2020). But last week, one of the Yenlo gurus actually came up with a brilliant story about visual recognition. Thanks to APIs, this concept enables a whole new way of collecting data. And that’s both useful and awesome.

It can’t be done

Since a couple of years, clever people are using APIs to collect all sorts of data in the outdoors. Think about weather conditions, traffic jams and license plate details. These data are collected, converted to analyzable information and then sent off to one or several databases for storage and re-use. However clever, those same IT professionals have been struggling to find a way to interpret data in photos, videos, and recordings. Visual and auditive information just seemed to abstract to capture, and their brave attempts either failed or resulted in inaccurate and complex tools. Most of the professionals gave up and admitted that it couldn’t be done.

Chances are this is a bridge

In a sense, they were right. You can’t expect a computer to tell you that you’re looking at a photo of the London Bridge, because a computer can’t think. What it can do, however, is learn how to recognize elements of bridges in general. This way, your computer compares thousands of photos containing bridges, until it is able to detect standard bridge elements such as water, sky, and handrails. Consequently, it will calculate the probability that there is indeed a tower in the photo. Your computer will get better at recognizing stuff over time, and provide you with more accurate information. This is done through so-called machine learning techniques.

Is that you, Elon Musk?

The awesome thing is, that the same goes for video and audio. You can actually teach a computer to analyze a video in terms of emotions, tone of voice, context, and the subjects that are being discussed. This is called Affective Computing and is not as new as you think. MIT’s Rosalind Picard wrote a great book about it in 1997! But as we’re now entering an era where new technologies allow us to optimize Affective Computing, it starts to become useable. And as computers gets more accurate over time, we’ll soon be able to analyze which person is in the video and what emotions they display. With a combination of face-, speech-, and context recognition, your computer will tell you exactly what you’re looking at and who you’re listening to. So, when some American guy starts talking about starting a community on Mars, you’ll know it’s Elon Musk.

Who’s working on this?

Let’s keep this article as readable as possible, and dive into some examples instead of technical details. One of our favorites is Clarifai, a young artificial intelligence company that offers an API that allows you to automatically tag an image with the elements in the picture without human interference. Any developer in the world can get an API key and join the organization remotely, while contributing to the database and using it him or herself.  Another example is Clarify (true story!) who takes visual recognition APIs one step further and specializes in audio and video. The company deploys deep learning techniques and traditional signal processing techniques to draw an image of what’s going on in a movie or a recording.

Digital snack

Lastly, Affectiva allows you to detect emotion, for instance through a webcam. You should definitely try this one yourself! Go to https://labs.affectiva.com/jssdk-fb-demo-video/  click on the arrow and watch the video of the pug being scratched. Make sure you’re visible to the webcam and your computer will tell you if you’re happy or not.

How visual recognition APIs are both useful and awesome-Affectiva's Demo for emotion recognitionFigure Affectiva's Demo for emotion recognition

What is it to me?

We may be innovation enthusiasts, we still work on solutions. In the end, technology should help you get the most out of your business rather than delight your IT department. So, let’s talk money. Visual recognition APIs unlock a treasury of data that would normally remain locked. With today’s technologies, you can analyze movies, radio shows, photos, and television programs and use this information to optimize your products and services. Now that you can go beyond written texts, you’ll learn so much more about customer behavior and preferences, which brings you closer to your potential and existing buyers. And as customer intimacy is an important parameter for success, visual recognition APIs are a huge business opportunity.

Visual recognition APIs are not just awesome: they lead the way to a whole new level of analytics. Still not convinced? Then think about all the hours you’ll save trying to sort your holiday snaps!

Looking for an API gateway or management tool to boost your business? Then download our API gateway selection guide below. It’s just as awesome as visual recognition- if not more awesome.

API Selection Guide 

Care to share?
   
Picture of Yenlo
Published April 11, 2017

Yenlo

Yenlo is Premier Certified Partner and Value-Added Reseller of WSO2. We're experts in API Gateways, Enterprise Integration and Identity and Access Management. Subscribe to our blog for the most recent updates!

Responses

Stay up to date with the latest articles