Inspiration

In 2005, 59% of the world resources were consumed by the wealthiest 10% of the population. In 2017 alone, enough rare earth materials were mined to give 45lbs to each adult in the US. Every week, New York City restaurants throw away 3.2 pounds of food per capita.

This waste epidemic prompted us to try to find a way to lower the barrier to entry for recycling. Many people seem recycling as a huge inconvenience - why take the time to figure out what type of material an item is made of to then have to spend more time figuring out if that material is recyclable, only to find out all the effort was for not and it ends up in the garbage bin?

With Wecyclr, that kind of tedious, manual effort is a thing of the past.

What it does

Using image recognition, our Wecyclr service identifies the materials that make up the object in question. From there, we apply a sorting algorithm of our own design to classify certain sets of materials into one of three categories: compostable, recyclable, or garbage. After the sorting is finished and the object is classified, we inform the user, whether through mobile app, web app, or voice assistant, of the object's category.

This mechanism is support in three different clients: an android application, and web application, and an Amazon Alexa voice skill.

How we built it

Leveraging the power of server-less infrastructure, we were able to use many services offered by Amazon Web Services to keep our codebase independent and manageable, with each service occupying a certain area of the product's features. A Go backend server powers the image recognition and text parsing, while DynamoDB was able to support any states that the services needed to be handled.

Challenges we ran into

Handling asynchronous REST requests with Alexa proved to be a challenge, but we are happy to report that the task was accomplished and Alexa is happily working with Wecyclr.

Accomplishments that we're proud of

Implementing 3 very different clients with a fully autonomous server in 24 hours is not an easy task. Our team worked like a well-oiled machine throughout the entire competition, and all were able to pick up a new skill or two along the way.

What we learned

  • REST requests with Amazon Alexa
  • Android networking
  • Object Detection via Computer Vision
  • various AWS services and features

What's next for Wecyclr

We would love to be able to implement our auto-recycler with a simple hardware element to really make this process automatic. Users would simply be able to single-stream all of their unwanted items and Wecyclr 2.0 would be able to physically sort them into the right bin for proper disposal.

Share this project:
×

Updates