Manipulation on Facial Attributes with Principal
Components Analysis
Abstract: Basically I am going to talk about facial attribute manipulation based on statistical approaches. Principal components analysis based facial model enables creating faces with different expressions. This muliple-creations as well as feature selection approaches help my research on finding relations between parameter selections and subjective comprehension on facial expressions, e.g., masculinity, friendliness and attractiveness.
Applications based on the above analysis include transforming a neutral face to be more masculine, friendly or attractive. Also I am going to introduce how to make a face more chubby or more skinny as well as how to perform facial expression transfers in the feature space.