Kanami Yamagishi, Takuya Kato, Ayano Kaneda and Shigeo Morishima

How Do Computers See Makeup?: Investigating the Effects of Makeup on 3D Facial Reconstruction

5th Women in Computer Vision Workshop in ECCV 2018, Posters

https://wicvworkshop.github.io/ECCV2018/

Makeup is a way to change one’s appearance. Previous studies on facial recognition show makeup’s potential to obfuscate one’s identity or to impersonate the identity of others. It is reasonable to assume that makeup would also affect 3D facial reconstruction because appearance matters a lot to both. In this paper, we investigated how computer vision sees makeup, and whether it is possible to infer the intention of makeup. We evaluated results of 3D reconstruction from images with and without makeup that intends to resemble targets, and images of targets. The result suggested that makeup affects 3D facial reconstruction and computer vision can capture the intention of makeup as it leads changes in facial geometry toward the impersonated faces. In addition, our evaluation found out that the effects of makeup vary depending on the part of the face. Our investigation revealed the impact of makeup on 3D facial reconstruction.