Trash/Recycling Image Detector

Inspiration

I chose this problem because I wanted to implement an incentive-based system in which users would be rewarded for recycling and helping to mitigate climate change. While I unfortunately lacked the time to complete the interface that would reflect this incentive-based point collection system, my original intention was to allow for greater recognition of how the public could be motivated to help improve the environment.

How I built it

I used transfer learning to develop a convolutional neural network model that recognizes uploaded images of various trash items and classifies them by the appropriate recycling class or as non-recyclable trash.

Challenges I ran into

Completing the web app was challenging, and I was only able to output a number for the class, rather than the appropriate trash/recycling class label.

Accomplishments that I'm proud of

In the full model, I reached an 82% image recognition accuracy.

What I learned

I learned how challenging developing the back end and front end of a web app can be. Although it currently does not function ideally to plan, I plan to continue debugging and improve the front end of the application, as well.

Built With

• keras
• python
• tensorflow