Inspiration

According the the United States EPA, the U.S. creates approximately 250 million tons of waste each year. For Canada, this is 31 million tons and for India this is 62 million tons. Not much of this waste is being recycled the correct way. Seeing this hazardous problem in our society, we decided to take action and create Picket. We wholly believe that humans have the capacity to come together, recycle their waste, and create a cleaner environment. We built Picket as a method to engage society in it's environmental affairs in the simplest way.

What it does

Picket allows the user to upload pictures of waste to the Picket website. Once our website receives the picture, it uses a machine learning model to correctly classify what type of waste the picture contains in 1 of 6 categories (Glass, Metal, Plastic, E-Waste, Paper, Organic). After it identifies what type of waste it is. It notifies the user to place that specific type of plastic into that categories' bin.

How we built it

We first used Flask to set up our server, and collaborated using VSCode's live share extension. We then added our google vision api to use as a machine learning model. We then grabbed the labels from the model and classified our test images with those labels.

Challenges we ran into

The first problem we ran into was installing numpy and some other useful libraries. Luckily we were able to fix this by restarting our live share. The next problem we were facing was that our google vision api was not accepting our image for some reason. We were able to fix this by looking of the documentation for some the os library we used. The next problem was that we had to change the names of some of our categories because google was not recognizing cardboard as cardboard as we intended so we had to filter out the labels given by the vision model.

Accomplishments that we're proud of

We're proud of our wesbite's UI design and functionality, and that we were able to implement a working google vision api that can scan images accurately.

What we learned

Some of us learned how to implement the google vision api because we haven't used the api before. We learn a lot about classification in machine learning and also the ways that classification can be used to benefit society.

What's next for Picket

The next steps for Picket is accessibilty. Implementing a mobile app with Picket as will as making it visually accessible with light and dark mode features is a step we not only deem necessary but beneficial for all of our users. We also intend to globalize our application but incorpating different language translations into our app/website! All the new updates we want to add is super exciting :)

Built With

Share this project:

Updates