Inspiration

We were talking to one of the hackers tonight and saw that people were investing in banks, but not investing in themselves so we set out to make an iOS app to change that forever.

What it does

It uses machine learning and mathematical algorithms to return a set compliments to a dish you take a picture of. After eating 6 plates of chicken last night, we were more than ready to take up this challenge. Assuming you ordered a salad at a restaurant, but you aren't very sure what usually goes well with that particular salad. The app will give you what statistically makes sense with a salad - for example, pasta.

How I built it

We pulled data from the indico api, used a python algorithm to quantify the most complimentary dishes based on 4000+ feature vectors, swift for the iOS app implementing a camera feature + a great UX platform, and a php server to store our data.

Challenges I ran into

With machine learning, the bigger the dataset, the better. Unfortunately, pulling data from API's such as Yelp yielded us 100x100 pixel images (wayyy too small) and pulling images from something like Instagram usually had some security to block us from downloading images with a script, so finding data was extremely hard and with 24 hours, it was a big roadbump for us.

Accomplishments that I'm proud of

We developed a UI that provides a simple and clean user experience. We coded a fully functional camera feature that saved new images in our database. Being able to use both php and python codes in unison was both a lesson and an accomplishment for us. Most of our members were also new to machine learning, so it was very challenging to implement algorithms and optimizing data types for our purposes. Being able to have an extremely accurate model is something we are all very proud of

What I learned

I learned the importances of use beta and alpha testing before using an API. Working with machine learning for the first time, we also found the freedoms as well as some limitations to the accuracy and effectiveness of some algorithms.

What's next for Feed The 6ix

  • Gathering additional data and pairing the image with restaurants to get more accurate and relevant dishes
  • Monetization with marketing efforts and accurate customer targeting
  • displaying health information for each food

Built With

Share this project:

Updates