On Friday night, we went to a restaurant in the city center in Laussanne. When we were reading the menu, we didn't understand anything. We had to translate all the sentences from the menu with google traductor. Also, we had to look for the currency value in order to know what was the real cost of the things, as we aren't used to swiss crowns. That's how we realized we coul do something to fix this and make our lives better, doing an app that with one photo gives you all the info you need! We can take a photo and the app can give you the translation of the menu, in your language and in the currency you are used to.

What it does

Taking a photo of a restaurant menu, our web app detects the different dishes and their price and gives you the translation of the menu in your language and in the currency you are used to.

How I built it

The back-end (where magic happens!) has been made in Python. We used two Google API's, the Vision one and the Translator. The first one was used to detect and process the text of the photo and the other one was used (obviously) to traduce the text obtained. The front-end was made with HTML, JavaScript and CSS. You can take or upload a picture and it is sent to the back-end to be processed.

Challenges I ran into

The most challenging thing we did was trying to correctly process all the different chunks of text of the photo in order to classify these ones in useless, name of dishes or prices.

Accomplishments that we are proud of

Most of us didn't know about Python and we hardly ever tried to use an API (a thing that was a little tedious to use correctly at first). Also we never had tried any sort of computer vision thing.

What I learned

We learned how to work with API's, Python, a little bit more of HTML and JavaScript and how computer vision works in a conceptual way.

What we would like to have achieved is recognizing perfectly the restaurant menus and prices, but we think that with our little experience on the field and the technology we have nowadays was a very challenging project to make it work perfectly.

Share this project: