Local Litter Reports retrieved using Firebase Storage
Login Page using Firebase Authentication
Succesful Object Detection from the Firebase MLkit, and succesful labeling from our own neural network
The neural network we trained is able to succesfully distinguish between litter and non litter objects it detects
Walking around our campus and city, we see trash discarded in almost all the places we interact with on a daily basis. For example, in the city of Atlanta and even on the Georgia Tech campus, it is not uncommon to find anywhere from a few pieces of litter to large mounds of it. This litter is not only unsightly and uncleanly but poses a significant danger to local wildlife and ecosystems, with plastics being particularly bad due to their tendency to degrade into harmful microplastics.
What it does
Littr incentivizes keeping the world around you clean by instituting a self-challenge and community challenge system. Littr serves as a crowdsourced database of litter sightings anywhere in the world. Whenever a user reports a piece of litter, Littr detects the type of trash it is and awards points to the user for helping keep track of litter in their community. The user may or may not decide to pick that trash up at that moment -- if they do, they are awarded even more points for their contributions, but if not, the litter will still appear in the Google Map API that tracks all of the litter in their community. Thus, someone else can come along and claim those points by cleaning up that litter.
Our hope with this project is to inspire individuals to take action against litter in multiple ways. The most obvious is direct incentivization, wherein the user can receive points and community leadership status for their contributions. In addition, the cumulative litter report map hopefully inspires users to address any litter either as soon as it is reported, or to help tackle an extremely large issue.
How we built it
We built Littr using Java in Android Studio to create an Android app with functionalities mainly derived from Google Cloud Products. Using Firebase, Firestore, Google Maps API, Firebase ML Kit, and Computer Vision, we created this lightweight, scalable, and powerful solution.
Challenges we ran into
Our first day of development, we had aimed to utilize Google's Flutter UI Toolkit as the backbone of our project, alongside the various GCP APIs we wanted to implement. However, due to the novelty of this technology, it was not meeting our needs and had to be scrapped in favor of base android, which ended up turning out well.
Accomplishments that we're proud of
We're proud to have created an app that has all of the functionality that we deemed necessary for such a project like a user authentication system, profile system, map API implementation, and camera/ML functionality. We even trained our own neural network solution to detect when an object was litter or not!
What we learned
We have learned an extensive amount about the structure of Android applications, especially their integration with cloud products and services like the Google Maps API.
What's next for Littr
Due to its scalable nature, Littr could have a lot in store for it in the future. Implementation of weekly community leaderboards for users to compete with others in cleaning their communities or perhaps sponsored events by local businesses could help drive up sales at the same time that the city around them is made better for all living creatures.
An interesting real-world use case for this app that stays true to the spirit of it would be if this were used by state or national park services to keep track of litter and other issues in the park, protecting the serenity of nature and the enjoyment of those who visit it.