What inspired us
The impending growth of climate change and its effects drives our group to take a step towards a green future. We've learned about people's lives being harmed across the world and knew actions must be taken. Climate change is super important to us, this is the 5th app that we've built to combat it, and speaking from experience there's no denying how threatening it is. When we started, we realized quickly how much carbon goes into carbon footprints of simple manufactured goods (over half a ton in total), we knew that we needed to spread the word, in a fast, easy, and impactful way. I remember visiting one of my favorite places in the world, Glacier national park (Montana), 2 times 6 years apart, and seeing horrifying changes, let alone the environmental impacts and impending famine, flooding, and mass-migrations that are coming.
What it does
Footprint is an app that allows the user to instantly track their carbon footprint simply by taking a photo of most objects. We provide you with an estimated mass of CO2 produced in mass released throughout the manufacturing process of your item. Scan a backpack, Scan your laptop, Scan your shoe! You will quickly get the carbon footprint of that item. On top of this, Footprint provides advice to users on how to aim to reduce their carbon footprints.
How we built it
We built this app using a combination of a webscraping algorithm that we built with the beautifulsoup framework and served on a flask server. We tried both a custom AI algorithm using the ImageNet (AlexNet) CNN and the firebase ML kit, which we ended up using just because of the ease of integration, even though accuracy dropped a bit. The app itself was built using flutter, a cross-platform iOS / Android platform that enables quick reload and software development.
Challenges I ran into
One of the biggest issues we ran into, was ironically, the wifi. Over 3 times, we spent large swaths of time thinking that we had severe bugs without code to find out that it was silly stuff like not being connected to the wifi, or a weird wifi connection error. Another big challenge was setting up a robust webscraping algorithm, we went through Chrome, Firefox, Safari, to end up using Bing as the main search engine.
Accomplishments that I'm proud of
We are proud of successfully setting up the webscraper, the algorithm took a lot of work. There was a lot of hyperparameter tuning and small algorithmic tweaks needed to get the machine learning functional as well. Technically we are very proud of this. More generally, we are proud of the impact we believe our product can create and the ability for good that it has. We think that the app can go on to seriously help others reduce their carbon footprint by making them realize just how much pollution their purchases are responsible.
What I learned
We had a couple of novices on our team, so they all learned the basics of flutter, app development, and teamwork. The more experienced members learned a lot about general web scraping, and since we're more used to building custom algorithms - CNNs, RNNs, etc., this was the first time that we really got any exposure to pre-made ml kits. Much more importantly, we learned that we create A LOT MORE pollution than what we ever imagined. Our laptops are responsible for between 300 - 450 kg of pollution! We have a lot of changes to make when we get home. :)
What's next for GreenFoot
We really want to build a more long term tracking feature (we didn't have enough time to make any in 10 hours), which would be part of a local database. This would enable users to total and modify more complete carbon profiles. We also wanted to incorporate additional footprints for things like cars/algorithms, there's this really accurate machine learning algorithm that we made some time back that we would really like to add to the app. The biggest thing for us though, will be getting our app into the real world, this means getting it onto app stores, marketing it to the world, and popularizing it. Social media embeddings can help spread the truth of how much we really pollute,