CSE498, Collaborative Design, Fall 2018
Computer Science and Engineering
Michigan State University

Whirlpool Corporation, headquartered in Benton Harbor, Michigan, is the world’s leading major home appliance company with approximately $21 billion in annual sales and 92,000 employees.

Whirlpool is reducing the daily challenge of getting meals on the table by providing their customers with a seamless kitchen experience. To this end, they recently acquired Yummly, the premier digital food platform.

Yummly gives personalized recipe recommendations based in part on the ingredients that a user already has on hand. Yummly determines these ingredients by applying computer vision to images captured with mobile devices.

In order to recognize ingredients accurately, Yummly compares captured images to a large set of existing images in which ingredients are already annotated with labels and validated for accuracy.

Our Image Recognition, Annotation and Validation (IRAV) mobile application leverages crowdsourcing of Whirlpool employees, utilizing users’ captured images and validating them, effectively contributing to the many comparison images required by Yummly.

IRAV users take pictures of ingredient items, annotate them, and submit them. However, before Yummly accepts an image, the annotations must be validated by other users.

An image is considered to be acceptable only after it is validated by enough IRAV users. Only then is it added to Yummly’s set of comparison images.

Our Image Recognition, Annotation and Validation application is written in Swift using Xcode for iOS, and Java using Android Studio for Android. Google Firebase is used for the backend.