Python Garbage Collection

Garbage collection but in real life and written in mostly python!

Overview

Our attempt at using a neural network to detect if an image contains trash (plastic bottles, paper/napkins, paper plates, wrappers, etc), common garbage found in classrooms and hallways.

Consists of a neural network training program and a small web app that allows users to upload images and check for items considered garbage, using our garbage finder and OpenCV's YOLO object detector. (Set to spoons, forks, bottles, and some other trash items.)

Network training: python_garbage_collection repository

Web app: trash-detector (Requires git-lfs)

Plans

Our dataset is, frankly, too small to make an accurate, robust classification network. A good source of image data with trash would help greatly.

Our image recognition code runs server-side, so in theory any app could send queries to a server running Python Garbage Collection and get info in real time. This would allow users to create littering-detecting cameras by sending frames from a video stream to our service.

Impact

This project has the potential to make the lives of janitors and students better. By alerting people to where trash is in real time, we can more effectively keep the school clean.

Built With

Share this project:

Updates