In the past few years, the attention span of the human being has dropped from 12 seconds to 8 seconds. When viewed from the perspective of the advertising industry, this change in attention span amounts to a waste of millions of dollars of lost revenue. Advertising plays a fundamental role in the dissemination of information for nearly every corporation and the overloading of advertising has served to the determent of all companies involved. In order to match with the present trends of the people,today's information spreading methods need to become more individualistic and less in the form of generic background noise the average consumer. In order to do so, we thought about improving something that everyone sees everyday, the digital billboard. Through the utility of the Watson API which is basically something which detects the type of things you're interested in based on you're appearance, age and gender in order to spread information that is relevant to individuals as well as supplementation by Google's Vision API, our billboards are able to further tailor and select advertising which is more impacful in their daily lives and thus help people gain knowledge about products they are more likely interested in.

What is Watson Advertisements

Watson Advertisement is an enhancement on the existing billboard by presenting relevant advertisements to people that are eyecatching providing a greater hit on the amount of the advertising which actually leads to an actionable result. The advertisement technique involved is using a camera which takes pictures of the area in front of the billboard at set intervals of time and uses the Watson API in order to categorize certain aspects of the visible and then change the billboard information into a relevant topic for a particular demographic. Watson advertisement only displays information that is relevant to the people that are standing in front of it and does not displays anything that is not relevant to the set of people around it. Furthermore, if a person is interested in a project, it is seen that the person looks at that particular advertisement for about 10 seconds and this billboard just need 8 seconds to know whether the person is interested or not based on whether if the person is present in the consecutive picture of the surrounding. If the person is present, then the billboard will display a video which is relevant to the image that was displayed earlier so as to slowly gather the attention of people.


A major problem what we faced was decreasing the latency from the stream into the Watson API as the camera must take pictures at certain intervals of time while being subject to a constantly changing view. The addition of latency makes it difficult to have responsive and immediate feedback to the viewer and cuts into the already short attention span. Our solution to this problem was to create a cloud server that could give lesser latency as well as subvert a technical limitation of the Nest Cameras inability to connect to Enterprise or Captive network through a laptop wifi passthrough.

What's next?

In order to make this a fully refined product, we would focus on using any USB camera in order to take the input that is able to distinguish in between the objects, have minimal latency, and hence can enable the Watson API to analyse images at a really fast rate and based on that display a more responsive which is suitable based on the type of people present around it. Advertisements to be displayed also have the potential of being downloaded directly from advertising partners who can update them at their own discretion. These billboards can be used in order to ensure that the company can push it's product to the consumer who are actually in need of it and the advertising done by the company becomes much more resourceful than it usually is.

Share this project: