During the last several years, facial recognition technology has seen many improvements within the field. Often, the research focuses on creating systems that generalize well to large groups of users via a one-size fits all model. However, there is the possibility that a higher accuracy could be achieved on a per-user basis by tailoring the model specifically to them. The goal of this project is to seamlessly integrate data collection and machine learning in a game to personalize a general model to individual users. A game provides a medium for data generation and motivates the user to provide authentic data by playing. Overall, our experiment shows promising results for personalization in games, as the personal model performs better than the general model in both speed and accuracy.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Hale, Samuel Alexander
• Racca, Stephanie R.
• Panetta, Allegra T.

Publisher

• Worcester Polytechnic Institute

Identifier

• E-project-040420-054210

Advisor

• Whitehill, Jacob Richard

Year

• 2020

Date created

• 2020-04-04

Resource type

• Major Qualifying Project

Major

• Computer Science

Rights statement

• In Copyright