Identa.Item

• 

  Classification model prediction

• 

  In-app interface

Inspiration

Right now, the market for fraudulent items, especially luxury items, is worth half a billion dollars. We wanted to see if we could solve part of this problem.

What it does

An app that lets you find that item -- a Shazam for products. Take a photo of a luxury purse and we will identify both the brand and the product

How I built it

• Python scripts for web scraping
• Microsoft Azure Custom Vision Services for ML
• Android phone application

Challenges I ran into

• Intricacies of web scraping and ML

Accomplishments that I'm proud of

• Learning a lot about web scraping and ML despite lack of experience

What I learned

• The many ways to web scrape with Python
• Using ML/AI/computer vision tools

What's next for Identa.Item

• Expanding to a greater number of products: first, to more brands and styles for purses, and then onwards to a much more diverse range of product types.
• Improvements to speed and scale of web scraping to increase dataset for training classification model
• Identify if item is real or fake and/or provide purchasing suggestions

Built With

• android
• azure
• python