Society has become increasingly aware of the long-term legacy of humanity on Earth. We want to support the next generation in making conscious consumer decisions. This is achieved by raising awareness with respect to the enormous carbon footprint of many ingredients that are shipped across the world or consume huge amounts of resources during production and help with a carbon reduced cocking experience.
What it does
Carbon Foodprint keeps track of your food pictures (on Instagram) and lets you know in case there is an easy way to improve your carbon foodprint. Pictures of meals are analyzed using a deep learning model and the ingredients and the recipe is jointly extracted. In case there are ingredients with a very high carbon footprint, the app suggests fitting alternatives with a smaller footprint. In addition, recipe instructions for a new and optimized meal are generated and the user can cook with the updated recipe.
How we built it
We're running a Python backend which regularly checks the user's Instagram profile for new posts. In case we detect a new post, the image is fetched and analyzed using one model that is trained to classify the meal, detects the potential ingredients and is also able to generate a recipe. In the next step, we match the ingredients to their respective estimated carbon footprints and also map ingredients to potential alternatives. The final step consists of letting the user know that we've found a good alternative by pushing out a push notification using Firebase Cloud Messaging and presenting the user the carbon saving alternatives.