Developing and Deploying a Jewelry Recommender System

The purpose of this project is to develop a jewelry recommender system using convolutional neural and other deep learning networks and productionize it by developing a web application using Flask

The presentation can be found in here: link

Prerequisites

All the required packages along with their version are in the requirements.txt. They can be easily installed with following command:

pip3 install -r requirements.txt

Required Steps

1. Developing a CNN multiclass image classifier to determine the jewelry type
2. Creating image clustering models (for each jewelry type) by using Autoencoder + KMeans. Autoencoder is used for nonlinear dimension reduction of images
3. Building a simple front end for the user interaction
4. Developing a web application using Flask module in Python
5. Connecting the front end and the web application by Jinja2 to get the data from user into the model and give back the results

Demo