FaceSearch uses the following approaches to find a look-like from an image.
FaceSearch uses our proprietary database of more than 300,000 people (100 images per person) to find closest look-likes from images. Such quantity and diversity of the data allows FaseSearch to find matches in an unprecedented accuracy and coverage never achieved by similar apps or services.
In addition to the large database, FaceSearch utilizes the state-of-the-art AI to search efficiently in the database and identify the closest matches to the image specified by a user. This process consists of several steps. First, th AI detects one or more faces from the image. Then, detected faces are converted to a high-dimension vector. This vector is also called features vector, as it effectively captures various features included in human faces. Lastly, the AI compares an obtained feature vector with ones from the database, and find the closest matches. Powered by our AI, FaceSearch can answer closest look-likes in an astonishingly short time for its quite large database of 30 million images of more than 300,000 people.