Keeping track of products and customers is one of the most common business problems. A concrete case is restaurant checkout, where a cashier selects the product that a customer wants to buy and the customer pays with card or cash. Often, this process of product selection and customer payment takes a long time, making queues arise.

What it does

AnyAI allows the business to record the current products being offered (e.g., on a daily basis) with a few image examples per product. Together with a userbase with profile pictures, AnyAI can visually detect the current product that a customer is buying and identify the customer at the same time. This way, the customer is charged automatically for the right product and checking out only takes 2 seconds, instead of the usual 10 or 20.

How we built it

  • Flask backend.
  • Fine-tuned deep learning model in TensorFlow.
  • React front-end capturing images from 2 cameras

Challenges we ran into

  • Accessing two cameras from within the browser.
  • Fine-tuning custom model.

Accomplishments that we're proud of

  • Easy to setup system: 5 minutes
  • Accurate

What's next for AnyAI

Deployment in University Restaurants

Built With

Share this project: