Trash.ai

Inspiration

We wanted to do something to help the environment and utilize technology for that end. One problem in waste management is that much of the municipal solid waste generated by humans goes directly to landfills without proper separation of waste streams leading to increased landfill size and greenhouse emissions. We found that there has been a significant amount of research in the automatic waste recognition using machine learning so we decided to try our own approach.

What it does

Trash.ai processes pictures of a single subject and returns if the item in the picture is trash, recyclable, compostable or not waste.

How we built it

We used Google's Vision API to label a picture and use the resulting labels to categorize the item in the picture. We created the model app for IOS using Swift.

Challenges we ran into

We discovered that the Google Vision client libraries for Java do not support Android which was an issue. Learning how to use the Google API's was also a challenge. We found categorizing a large number of different labels to be difficult.

Accomplishments that we're proud of

We're proud to have created an application that works and learned how to use an unfamiliar API. Much of the time was spent learning how to use the tools necessary to build the application.

What we learned

We are helpless without Internet.

What's next for Trash.ai

The original motivation behind Trash.ai was to create a robot that could clean up the streets by picking up trash and correctly sorting it. Thus, the next step would be to implement this service in a robot that would automatically identify different types of waste on the streets and pick the waste up.

Built With

• swift-google-vision-google-cloud