camera object recognition shopping webapp

When the hackathon started, our team only wanted to try and implement a working object identification system. Overtime, as we tried various current implementations, we wanted to do something more useful.

When someone mention Amazon and its amazing cahierless stores, we finally decided on building an extensive shopping checkout system on top of the object identification system.

We settled with the tiny-yolo system for its optimized performance and ability to run on a client's browser.

We created a small look up manager to store the item names and prices, then we connected it to the objects templates name that the current model can identify.

Then we made a simple user interface: an add item button to add the current recognized items, an item list to show the current items to be checked out, and a checkout button to checkout.

Then a remove item button was made because we know the framework will make mistakes.

Now the framework can effectively identify sale items (items not listed will not be added).

Our next steps is to add a receipts page and a profile page to show product recommendations, We also plan to create a server back end to keep the prices updated and to add new item definitions.

Built With

• css
• deep-learning
• html
• javascript
• node.js
• yolo