What did you look like at eight? What will you look like at eighty?
Try out our "face ageing" system in person. No fancy make-up required! Our system takes a live image, and uses biometric and machine learning to extrapolate what your face looked like when you were younger, and what it could look like when you are older.
The system has potential to help authorities find images of criminals when they only have old mugshots, or to complement primary biometrics in border security, or to help forensics locate missing people.
The technical challenge is that this kind of projection and matching is sensitive to facial alignment, partial obstruction of a face, and low quality images, as well as facial aging. That's a complex process that affects both the shape and texture of a face.
Our progression system can make an impact on verification accuracy, with novel applications such as matching driver’s licence, passport, and visa images with the real person’s images.
The system has designed, prototyped, and evaluated a new method for age progression that handles misalignments and different types of noise. The method treats a face image as the combination of an age-specific component and a specific component representing expression, shape, and identity.
Machine Learning Classifiers are trained to extract the age specific elements from a large database of face images, so that an individual subject image presented to the system can be manipulated warped to represent the same face at a range of different ages.
Are you going to grow up to look like your parents? Come and face up to the possibility.
This is one of the demos in our Emerging Technology Lab Tour.