Recycling classifier

OnRender will take a minute to load for free tier projects

Inspiration

We wanted to make a waste/recycling classifier because we've often had trouble deciding what item should go into which category. When waste is improperly sorted, garbage can end up being mixed in with recycling, rendering the load unrecyclable. In cities like Toronto with high contamination rates, the contamination rates can reach as high as 25% of the recycling.

What it does

The website is a simple interface where you can upload a photo of your item, and it will return the model's best estimation of how you should dispose of that item.

How we built it

We built the project using Flask for routing and file handling, and we used the Roboflow API to train a model to classify the images. We took our training dataset from https://www.kaggle.com/datasets/mostafaabla/garbage-classification

Challenges we ran into

We were unfamiliar with any technologies we wanted to use, and we struggled with trying out various technologies and frameworks such as Svelte, Vue, PyTorch, and TensorFlow for the first time without making any progress at all. Later at night on the second day we found the RoboFlow API, and we decided to try to make something a lot simpler with just Flask and vanilla HTML/CSS

Accomplishments that we're proud of

Making a project while using Flask and Image Classification Machine Learning for the first time!

What we learned

We learned the basics of Flask routing, and discovered the Roboflow API.

What's next for Recycling classifier

We'd like to improve the front-end and add more resources that users can look at

Built With

• flask
• roboflow