FASTER FACE CHANGING TECH

Abstract

Nowadays, “Face Recognition” was introduced to our daily life, it has been popular for people who are interested in computer vision. Among different applications, “Face Changing” is one of the hottest topics, but most of such technologies are not as good as we thought. Also, difficulty to implement is another issue that results in unpredictable outcomes. In this project, our goal is to improve the similarity and generalizability of face changing by adjusting existing algorithms, making the whole process much easier to handle. As an ultimate goal, we hope to get a better output compared with original image. First, we have observed that the original algorithms only allow similar face angle to get somewhat good result. As a result, we try to focus on this point. However, we encounter some difficulties. Second, although the original two pictures have similar face angle, it still has some combination problems.