README.md

ecoshopper ♻️ 🛒

Ecoshopper was a prototype mobile web application designed as part of a final project for the Summer 2021 iteration of UC Berkeley's CS 160. The core of the app is an image classifier (pretrained VGG-16) that was fine tuned on the TrashNet dataset for the task of identifying recylable goods from non-recyclable goods. The model pipeline involved a fairly unique method that extracted metadata from an item's barcode to reverse search n (n being odd) images most relevant to the query. The classifier ran inference on the images and returned a prediction (recyclable or non-recyclable). The model was implemented with PyTorch and deployed with Django, while the front-end was built with React Native.

Note: This repository is no longer active.

Installation & Running the App

To reproduce the environment and necessary dependencies for this project, follow these instructions:

1. First, make sure you have npm installed. Follow the guide here: node.js Download

2. Change into the ecoShopperReact folder with:

   cd ecoShopperReact
   

3. Run the following commands:

   npm install -g expo-cli
   expo install
   expo start
   

Then, the Expo developer tools should open in your browser window.

4. Next, click "Run in web browser". This will open up a new tab. After, access the developer tools on your browser and change the device layout to a mobile layout (e.g. iPhone X). Additionally, the app can be run within the Expo Go app on a mobile device.