HackDavis 2018 to improve on Google Shopping by matching one step further.
- Aggregate all shopping products from different companies onto one site.
- Allow user to look for more specific items based on selected images (which are closer to what they are looking for).
- Creating a model that will successfully and accurately assess similarity between 2 images.
- Get enough quality product data (professionally taken images, product links, product title)
- Use NLP on product titles as another layer to assess similarity between products
- Train different models (convolutional autoencoder?), different loss functions, and different arcitectures for best results.
- Get more data.
- Tensorflow + Keras
- Google Cloud
- Python, Flask
- SQLite
- Javascript + JQuery
Coded by Varun Ved & me. (+1 person last minute to help with front-end)