Inspiration In an era marked by the escalating climate crisis, we sought to empower individuals to make sustainable choices, specifically regarding their food consumption. The lack of accessible and easy-to-understand information on product's carbon footprints sparked the idea for co2mpare.
What it does co2mpare is a mobile application that calculates the carbon footprint of food products by simply scanning the ingredients list. It provides real-time CO2 emission data, enabling users to compare products and choose the more environmentally-friendly option.
How we built it We built the main application using Swift, creating a user-friendly UI for seamless user interaction. Image recognition was facilitated by leveraging Apple's CoreML for converting food labels into machine-readable data. OpenAI's embeddings and cosine similarity enabled the efficient matching of this data with our CO2 emission database.
Challenges we ran into We grappled with making the image recognition model efficient and accurate enough to handle different fonts, handwriting styles, and languages. Enhancing the accuracy of CO2 emission data retrieval was another challenge, particularly when dealing with complex and ambiguous ingredient lists.
Accomplishments that we're proud of We're proud to have successfully built a functional app that can calculate and present a product’s carbon footprint in real-time. In doing so, we hope to have made a valuable contribution to individual and collective efforts towards sustainability.
What we learned Through this journey, we gained a deeper understanding of Swift programming, machine learning models for image recognition, and the practical applications of cosine similarity in data retrieval. We also learned about the significance of embedding technologies in enhancing algorithm efficiency.
What's next for co2mpare We plan to expand co2mpare's capabilities by incorporating additional features, like tracking a user's carbon footprint over time. We also envision scaling our database to include a wider range of products and making the app multilingual to reach a global audience.
Log in or sign up for Devpost to join the conversation.