Inspiration
The group was curious about the image recognition API provided by Clarifai, and decided to build a project around it. The hotdog/not hotdog app from the TV show Silicon Valley came up, and we were all excited about the idea of building an app that provides information to people with a picture of food. After some discussion, the team decided that it would be a good idea to inform people about good drinks that can be paired with the food.
What it does
Foodify can take a picture of a food, and suggest a drink that goes well with it.
How we built it
We decided to build an Android app with Java. We divided up the tasks, one person on GUI design, one on interaction with Clarifai's API, one on building a database for food-drink pairings, one on basic app logic, and one on writing the logic for suggesting drinks based on the kinds of ingredients recognized. We developed our parts independently and integrated it later.
Challenges we ran into
The main challenge is to build the database for food-drink pairings. Time constraints doesn't allow creating an actual algorithm that pairs food to the perfect drink. Our solution was to store data as simple key-value pairs of food names and drink names, based on our research from Google, and use a weighted average of the drink suggestion for every ingredient recognized from Clarifai to recommend a drink. However, the result from that is not accurate enough.
What we learned
Most of the team didn't know android development, and this has been a very valuable experience.
What's next for Foodify
If time permits, we would like to build a more detailed database of food and drinks, as well as a more accurate algorithm for pairing them
Log in or sign up for Devpost to join the conversation.