The idea came from experience in a fine dine restaurant. Everytime I go there I get confused looking at the menu. I try to read the name and visualize the dishes but the names just dont make any sense to me. So I call the server and start asking him about the dishes one by one or I google search for images of the dishes before making a choice. This got me into thinking what if I dont have to manually search each item or disturb the server so many times.
What it does
It uses OCR technology to augment the menu and scan the text(dishes). Then it gives information such as images, country of origin of the dish and ingredients of the dish.
How I built it
Foodie is an android app that uses Google's OCR packages to detect text among images available on Mobile Vision. The application fetches information about the dishes by querying the back-end. Back end is made using Flask and Google Knowledge Graph API and Duck Duck Go API.
Challenges I ran into
Google's OCR packages are complex and it took some time to understand how those packages work. Once information is read from stream of camera frames, the next challenge was to process text and send it to the server. We spent some time figuring out how to send cross-domain requests. Finding the right API was a challenge as most of the search platforms are not open source. There was a steep learning curve for Python-Flask application framework.
Accomplishments that I'm proud of
Lots' of learning was involved be it Google's OCR technology, Python-Flask framework and back-end Duck Duck Go API. We are glad to complete the app in the given time.
What I learned
Google's OCR technology, Python-Flask, Duck Duck Go API but most importantly time management, team work, spontaneity and creativity !
What's next for Foodie
Proposal: It would be nice to integrate Foodie into wearable devices such as Google Glass and see where we go from there...