own_challenge_SNAX

User input
Mockup after taking photo of grocery
mockup after taking photo of grocery

Inspiration

Since two members of our team are into body building healthy nutrition has been always important. The need to track items one by one in apps by scanning QR-Codes is exhaustive, time-consuming and tends to get neglected after some time. App with possibilities to scan QR-Codes of items while grocery shopping is extremely slow if you have multiple items and furthermore lacks an aggregated overview of all items in one single view. This is where our idea for this project was born. Beside improving the Customer Experience of grocery shopping, we want to help people provide a healthier lifestyle by also tackling malnutrition and consequently Diabetes 2 and other diseases. ...

What it does

Initially when opening the app the user gives inputs on his personal body information (height, weight, daily activity) based on that the recommended calory intake per day will be registered. When going grocery shopping the user takes a photo of his shopping cart. This image is processed and the selected items will be detected and the nutritional macro values queried from the list of all store items. In the next screen the user will see the cumulative values of all important macros (as bar charts) such as carbohydrates, protein, sugar and fat for his grocery cart items. Whenever a value (f. ex fat) reaches a critical value an alert will be given, and a recommendation for replacing one of the unhealthy items with an healthier alternative will be given.

How we built it

Verim: Strategy and Product Development Yangtao: Strategy and Product Development Ajay_Input: App Development Qiao_Input: Image Recognition/ML Model Eva_Input: Dashboard

Challenges we ran into

..were mainly in the object detection area. Finding useful datasets which include items from the grocery store of "Rewe" was difficult. Furthermore, the time constraint was a problem for the training of the model. (Less than 24 hours were left are setting everything up for this part). Besides that, we also faced a generalization problem of our real data to the test data. (Detection good working on data set but not real life items) ....Training resources (Setup GPU on laptop) Another (minor) challenge was faced on deciding which nutritional information the average user needs to see in his app and which one can run in the background. - UX/CEX

Accomplishments that we're proud of

We are especially proud that since the Kick-Off-Event the team harmonized pretty well. After going through the goal of our app and the major milestones everybody worked intensively on putting his/her expertise into work.

What we learned

We learned that tough or nearly impossible time limits challenge most of us personally and on a team level. In fact, this helped us to come up with innovative ideas on how to constantly face challenges adequately.

Furthermore, we also learned from each other's expertise.

What's next for own_challenge_SNAX

Continue our work :) Work with the first big grocery store partner together -> get their data for improving the ML model Partner up with doctors to make the app more personalized regarding needed nutritions.