In this project an web app has been created to find similar looking amazon products. This can help find the best deal.
The app users can given a product link and similar products can be found. The app uses deep learning method to compare images of product.
The model contain a CNN model that find embedding for product images and then find the nearest neighbour using KNN . For training the model the kaggle competition dataset shopee-product-matching was used.
The model is deployed in AWS
Demo video of App: https://youtu.be/16IZpM6QmW8
Link to Web App : http://35.154.205.146:8501/
Note the web app is not working as amazon is blocking the ip address , hence unable to scrape product. Solution is to use Amazon API to fetch data.
1> pip install -r requirment.txt to install required libraries
2> Run python download.py to download the pretrained model
3> streamlit run app.py