For Snack Overflow/Benson Store items that lack barcodes, customers must navigate menus to select desired items. To make this process more seamless, we propose to use image recognition to recognize these items to make up for these gaps in barcodes. Users then must only confirm the recognized item. Furthermore, when items are identified incorrectly, users' later selection of the correct item provides additional labeled data to train the image recognition model, providing a solution that improves with time.
- Bananas
- Tea bags
- Muffins
- Cold Brew Coffees
We heavily relied on a UW CSE CSE 455 homework project for structuring the PyTorch-related code.