The present billboards are passive in nature, the display of advertisements on such billboards leads to unstable profit generation and moreover, advertisers don't get the relevant audience for promotions.

What it does

Our Digiboard is a B2B solution involving the promotion space provider and the advertisers. It uses a camera on the billboards that makes use of Azure Cognitive Services VISION API to generate demographics of viewing audience that can be used to predict demographic trends every fifteen minutes. Advertisements are then matched to the billboards depending upon billboard with the highest relevant audience based on queueing based matching algorithm.

How we built it

We have used an Android application to capture image frames from real-time video and send it to our backend for processing. We plan to use high-end cameras specialised for image processing in future. On the backend, we get the demographics using Azure Cognitive Services. The queuing based matching algorithm was written at the backend which matches advertisements to the billboards. For demonstration purpose, we have also built an AR application to showcase the view inside the mall with a sample of 4 billboards and sample data to simulate the matching process. An advertiser portal has been made to track the advertisement progress. Tech stack used is - Javascript for the front end, PHP for the backend, C# for AR application.

Challenges we ran into

We found it difficult to achieve good accuracy while detecting faces in front of our billboard using haar cascades in python. Then we decided to use Azure Cognitive Services Vision API to do the same process which has a higher accuracy and we were able to achieve good outputs.

Accomplishments that we're proud of

The benefit of Digiboard is that the pricing is based on relevant viewership thus acting as a boon for both advertisers and the promoting space providers. It can be scaled up for usage in malls, restaurants and airports.

Built With

Share this project: