NutriFence

ArchHacks 2016: Winner of the Best Nutritional Hack by Panera

Matthew Watt, Tanay Parikh, Isabelle Sauve, Daihan Zhu

Devpost: https://devpost.com/software/nutrifence

Core Features:

• NutriFence is an iOS app that allows you to take an image of a list of ingredients and have it tell you whether or not that item is safe for consumption if you have Celiac's
• Core requirements:
  • Capture the image & format it correctly
  • Send this image to Processing API (OCR)
  • Send the OCR results to the Classification API
  • Perform a spell check on these results
  • Recieve and display information on the safety of the ingredients

Components

Database

• Redis (Mac): http://redis.io/download
• Redis (Windows): https://github.com/MSOpenTech/redis/releases

OCR

• Google Cloud Vision API: https://cloud.google.com/vision/docs/

Spell Check

• Bing Spell Check API: https://www.microsoft.com/cognitive-services/en-us/bing-spell-check-api

Classification API

Requirements

• Recieve a POST request in JSON format from the app
• Spell check this data
• Compare this information to a database of "unsafe" items
• Determine which ingredients are 'safe', which are 'unsafe', which are 'risky' and send these back to the app
• Draw conclusions based on extracted data and send this back to the app
  • is it definitely not gluten-free?
  • are there contains that indicate it might not be gluten-free?
  • does the item actually say "may contain wheat" on it?

Implementation

• uses Redis as a data store for storing the categorization of ingredients
• uses ExpressJS and Node.js to recieve and handle POST request
• Reformat the raw data, including sending it out to the spell check API
• pulls lists of unsafe/unfriendly ingredients from Redis, then parses the request
• execute algorithm for cleaning up the raw data recieved
• sort every item into appropriate category by comparing the ingredients in the request with the lists of ingredients from Redis
• return data & conclusions in JSON format back to the app (POST reply)

Configuration

Create environment values

1. Add a new file named .env to /Classification API/.
2. Add the following values:

BING_SPELLCHECK_API_KEY=***SPELLCHECK-KEY***
GOOGLE_VISION_API_KEY=***VISION-KEY***

Installation

$ npm install

$ npm start

Use

TypeScript autocompilation & reload:

$ gulp