Many people are realizing that climate change is real and that our daily choices have a direct impact on our future. Food production is responsible for approximately one-quarter of the world's GHG emissions so people willing to reduce their carbon footprint are becoming more open to trying a more sustainable diet. But, what food is really better for our environment? And can there be a simple tool that can help us make a more informed decision?
See also Challenge #1, 2 and 18
What it does
By combining Migros huge recipe database Migusto with Eaternity, the largest environmental database worldwide, we set out to build an app which helps a person visualize the environmental impact of each meal in a restaurant. In this way, he/she can make an informed decision and reduce his/her carbon footprint, with only a few taps on a phone.
How it works
Using our app, the user takes a picture of the menu with his/her phone. The app converts the photo into text and, using the recipe database as well as the Eaternity impact scores, it displays an environmental rating for each menu option. Additionally, information about the CO2 emitted, the water, forest, animal and seasonal impact is shown. This way, the user can have a better overview of the impact of that dish and make a more climate-friendly decision.
How we built it
In order to follow a user-centered design, we started by defining a persona. We gave this person a name, an age and a job. Then, we defined the Empathy Map of what this persona feels, thinks, says, does. After, we identified the pains and the journey/steps that this person takes all the way to the point of making a decision on what to order in a restaurant. Based on all this, we started thinking about how the app should function, it’s look and feel.
On the technical side we use several OCR and computer vision techniques to de-noise the image and extract menu names. The menus are then passed to our Python backend hosted on Azure with an automatic CI/CD pipleline in place. In the backend, we prepare a customized query for the Elasticsearch cluster of Migusto so we can match the menus to their corresponding recipes as best as possible. Once we have the recipe we can extract all ingredients of a given dish and pattern match it with the products in the Eaternity database. We can then use the Eaternity API to get a wide variety of environmental indicators which we present the user in a cross-platform mobile app.