DigiMart

Digital browsing for an analog supermarket

Comment 1

Inspiration

Supermarkets get larger and more complex and customers are met with dozends of brands and product variations. So paying attention to keeping a healthy diet and choosing the correct options can be quite overwhelming. Its time to stop wasting half an hour checking the back of all products to find what you're looking for.

What it does

DigiMart allows users to tag and color-code products depending varying requirements.

Do you want to search for a specific product? Or maybe choose low-fat options without checking the whole row? You may also color-code the whole rack to see the relative price of all items in a range from green to red. Or be warned not be take products that contain ingredients that you're allergic to.

These filtering modes can be chosen in the app, can be extended with further functionality and new modes can be defined by us as well as third-parties to fulfill every customer's wishes.

How we built it

The product is built as an iOS application using Apple's ARKit, an AR toolkit for the Unity engine. Using this framework as well as planar image recognition we detect the products that have previously been fed into a product database also containing all product data and nutritional information. The different modes and their color-coding are implemented with a strategy design pattern working on the data bag of a single recognized product. Therefore new modes can be defined easily.

Challenges we ran into

ARKit has certain limitations regarding the performance of placing overlays and recognizing products. We have thought about possible optimizations using alternative methods but sadly the time is not sufficient for those. Also we only have a single Mac in our team, limiting the work that can be done in parallel on the AR-aspect of the app.

Accomplishments that we're proud of

The product recognition worked much better than expected and were proud with the solution we found for the filtering modes. We were also able to define a full-text product search as well as animated advertisements on preselected products.

What we learned

AR has advanced a lot and only lacks real-life applications so far. Hopefully DigiMart is a first step for society to accept a digital companion in offline situations and the way they can extend the everyday experience. There are many more possible applications and we are happy about our new-found knowledge in AR-related tech.

What's next for DigiMart

DigiMart collects data that can be used for much more than just color-coding. Using the continuous image stream collected over multiple users in the same location we hope to be able to construct a 3D map of the supermarket to be able to also guide users to the product they are looking for.

Built With

Share this project:

Updates