TORec

Inspiration

There was one day where my roommate and I were arguing about where to dispose a takeout container and to be honest, we were kinda lazy to search that up or go to the Toronto website and look for their tool which would take quite long. So I thought, wouldn't it be cool to have an app that makes the process of recycling and disposing garbage much easier and more convenient. That's how I came up with TORec.

What it does

TORec will allow users to scan their waste with phone camera and then show users the proper instruction of disposing

How we built it

I use react native for front end development. I trained the image classification model with Tensorflow Object Detection API and OpenCV

Challenges we ran into

One of the challenges I ran into was my laptop GPU was not able to accommodate the Tensorflow model. For that reason, I'm still working on changing some of the architectures to make it more light-weight. The Tensorflow model unfortunately won't compile so I only have a MVP.

Accomplishments that we're proud of

What we learned

I really thought deep learning was super complex and difficult to learn due to the math parts behind it. However, the software is very well-documented and there are a lot of youtube tutorials that teach some core concepts which is enough for me to leverage and apply into building TORec.

What's next for TORec

Hopefully TORec will have a better object identification algorithm that outputs better accuracy.

Built With

• opencv
• python
• react-native
• tensorflow