Estimate position of lower facial features from upper face
We use upper face image containing forehead, eyes and upper nose bridge in the input image to estimate positions of lower facial features like mouth, nose tip and jawline. This position estimation is carried out using face annotations (landmark points). Dataset used is Helen Dataset (http://www.ifp.illinois.edu/~vuongle2/helen/). It consists of a total of 2330 images having 198 annotations per image.
Sample Outputs:
