MMER_FEASy - the FacE Analysis System

All know-how and technologies of MMER-Systems are aggregated in MMER_FEASy, the FacE Analysis System. It makes use of all Components

• MMER_Lab,
• MMER_GPU, and
• MMER_Locate

and combines them to a system which fully automatically extracts the relevant information from human faces.

MMER_FEASy consist of the following MMER_Lab modules:

• Image / Video / Camera Source:
  Various types of input can be loaded and are provided to the subsequent processing chain.
• Head and Eye Tracker:
  The MMER_Locate technology detects the head in the current image or video frame. The localization of the eyes ensures a reliable initialisation of the following Face Analysis.
• Face Analysis:
  This is the heart of MMER_FEASy. Based on an enhanced Active Appearance Model methodology, the face to be analysed is reconstructed by adjusting a number of parameters describing the "appearance", i.e. the hull of the face, the shape of facial featues, and their texture. These parameters carry all relevant information necessary to subsequent evaluative modules.
• Classification (optional):
  Based on statistical models derived from large databases, various labels can be assigned to the current face or the analysed video sequence, e.g. gender, age, facial expression.
• Avatarisation (optional):
  Since the hull and shape of the face and its facial features are reconstructed during the analysis, this information can be applied to any photo of other faces such as celebrities, aliens or other human-like and artificial avatars.

Please visit Application Areas to learn more about possible applications.
Do you have further ideas? Please contact us directly...