How Watson & Bluemix Helped Bring Our Ad Targeting Idea to Life

Share this post:

How crazy is it that a simple thought or thread of a conversation can develop enough momentum to change your life? We are a group of four friends and by profession, we are software developers. We have been working in the services industry for the past five years and programming has been our passion.

A couple of years back, while a friend of mine was waiting for his flight at the Bangalore airport, he realized that most of the advertisements placed at the airport were not very relevant for the travelers. The question it sparked was – is there a way to help marketers analyze the presence of target audiences, their interests before putting up an ad in a particular area or geography?

The four of us brainstormed and realized that this could be done through the concept of “computer vision.” Though it is a very broad concept, we used it as the base to develop a solution. Computer vision is a field that includes acquiring, processing, analyzing, and understanding images. In addition, it deciphers high-dimensional data (image parameters ranging from brightness to pixel density) from the real world to produce impactful insights in numerical or symbolic form.

So we built an app called Clairvoyance that automatically plays advertisements on display boards based on the age and gender of people in the vicinity, as well as the location of the digital display board. In the future, we want to add to the app the ability to measure audience reactions to the ads.

We built the first prototype using Microsoft Kinect, where the application would train itself by learning faces by capturing pictures and tagging them appropriately. The next step was to extract the interests of each individual from their social profiles. This information would then be used to display appropriate advertisements by using Kinect to capture pictures of people standing closer to the ad.

While the solution worked, it was not scalable, cost-effective or easy to deploy. A couple of years later, we came across IBM Watson APIs that were made available for developers over IBM Bluemix. Watson APIs enabled us to do a lot of things that used to leave us struggling in the past, such as text to speech, question and answer API, summarization etc.

Not only were these API’s easy to consume, but also reliable and accurate. We found a good set of Face recognition and detection APIs that helped us to segregate data such as age and gender of the people recognized in the image. (Click here to view the API bundle that we utilized.)

Bluemix provided us with a host of other services such as NoSQL Storage, Integration with Build tools and source controls as well. This easy-to-use development environment enabled us to build an application that was robust, scalable and cost effective. Previously, while developing the app using Kinect, most of our efforts and time was consumed in getting things done (Development, Testing etc.). With Bluemix and Watson, we could focus and brainstorm on marketing our idea more than dealing with operational challenges.

With the support from various hack fests, the application has garnered a lot of traction. While we have not yet launched the app commercially, we are currently working with two shopping malls.

Apart from just playing advertisements, the information that we capture and store in our systems could be utilized in multiple forms for our clients and we will keep things posted soon on our website http://www.3dot2.com. To sum it up, the past six months have been a roller coaster ride for us. This journey wouldn’t be possible without Watson and Bluemix which truly helped us bring our idea to life!