CarbonSee

Inspiration

As a team of international students, we fly a lot, so we are conscious of our carbon footprint and on the hunt for ways to improve it. The first step to improvement is awareness, and so we decided to tackle the task of making one aware of their carbon footprint and its contributing behaviors.

What it does

After buying your groceries, simply take a photo of your receipt with our app. We extract the details of every item you’ve bought, and then use machine learning to match it to our database of grocery items and their carbon footprints.

The results are conveniently added up and displayed to show you your carbon footprint, and the proportions of items that contribute to it.

This gives you a sense of which items you buy contribute the most and least to climate change, helping inform your actions to lessen your climate impact.

How we built it

Our app is built using the Huawei HMS core. We use the API to extract text from the image. The text is then scanned for keywords and parsed, removing useless punctuation and noise. Next, we use word embeddings to match items in the receipt to items in our Carbon Footprint Database.

Accomplishments that we're proud of

1) Being able to successfully use the Huawei APIs for Text Recognition and Text Embeddings. This was a major challenge because of our lack of experience with these APIs, however, we soon found them intuitive and easy to deploy.

2) Integrating Python code for data analysis and visualization with the code for Android Studio. We managed to link these 2 disparate components using Google Firebase, capitalising on its ability to store databases in an accessible, fast, and lightweight manner.

What we learned

1) Basics of developing apps with Android Studio.

What's next for CarbonSee

-> We can recommend users to swap certain items in their diet with suitable alternatives. -> For example: Replacing a beef heavy diet with a pork heavy diet -> We can track people habits over time, and give them positive reinforcement for positively changing their habits -> Introduce a social aspect to this, where people can see how their friends and family score on their carbon emissions IQ score: Do they choose more eco-friendly products?