There is a dearth of research literature investigating the effects of cosmetics and beautification on the accuracy face recognition technology. This technology is being deployed more frequently as a means of unlocking smart phones, gaining access to financial information/buildings and logging into computing resources. There have been some studies that have found that these algorithms don’t perform as well on women. A longstanding question has been does the application of cosmetics cause the degradation in performance/accuracy? This research project attempts characterize/assess the impact of beautification and cosmetic applications on the accuracy of biometric face recognition.
To conduct a study on women, cosmetics, and face recognition, a huge part of the experiment is to collect the images of women over the Internet by creating a web portal which would allow women to upload multiple selfies to support this research:
1. No Makeup – an image where a person doesn’t have any makeup
2. Light Makeup – an image where a person does have very light amount of makeup (ex. foundation and a blush)
3. Heavy Makeup – an image where a person has full makeup (foundation, contouring, lipstick, blush, mascara, eyeshadow, etc.)
Collected images will be tested to assess whether or not the application of cosmetics reduce the performance/accuracy of facial recognition.