Our main source of information was the effort to keep our cities clean in an age where waste generation is reaching higher levels every year. We realised we have to increase the recycling effort at a societal level and provide people with tools to help that happen at a micro level.
What it does
Our application categorises items that the user wants to dispose of. The user can be a line in a recycling facility, redirecting items to their correct recycling process line, or a simple user that wants to get on with their lives.
How we built it
We have used Google Cloud's Vision API to classify images of objects that have to be disposed of. The API provides a label set which is used to assign the item to one of the recycling categories enforced by the council's recycling policy. The project also has a simple and easy to use web application to allow users to scan items from their phone wherever they happen to be.
Challenges we ran into
Classifying items from the label set provided by the Vision API proved to be a challenge since it required a lot more NLP techniques than expected to come to satisfactory results.
Accomplishments that we're proud of
We consider that a great achievement of our project is successfully integrating Computer Vision and NLP APIs to create a simple to use application that could save our users time and effort.
What we learned
Most of the technologies we used were new to the members of the team, and we consider we have learned a fair bit of Flask and computer vision.
What's next for CleanCity
The idea of using digital solutions to improve recycling amongst an internet connecting audience can be further improved. One idea that the team is thinking of implementing is instantly presenting users with a map of recycling bins closest to them, filtered by category, and useful insights into how much recycling helps reduce pollution in and around their cities.