Illumination-Aware Age Progression

CVPR 2014

Ira Kemelmacher-Shlizerman, Supasorn Suwajanakorn, Steven M. Seitz


Abstract

We present an approach that takes a single photograph of a child as input and automatically produces a series of age-progressed outputs between 1 and 80 years of age, accounting for pose, expression, and illumination. Leveraging thousands of photos of children and adults at many ages from the Internet, we first show how to compute average image subspaces that are pixel-to-pixel aligned and model variable lighting. These averages depict a prototype man and woman aging from 0 to 80, under any desired illumination, and capture the differences in shape and texture between ages. Applying these differences to a new photo yields an age progressed result. Contributions include relightable age subspaces, a novel technique for subspace-to-subspace alignment, and the most extensive evaluation of age progression techniques in the literature.



Files

Paper, Additional results, YouTube video, FGNET dataset
IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), June, 2014.

Results

Below are many examples of our automatic age progression. For each example, we show the input (left) and automatically "aged" version of the input using our method (right). Age is specified below each photo.

Males:

Females:

Comparison

A single photo of a child (far left) is age progressed (left in each pair) and compared to photos of the same person at the corresponding age (right in each pair). The age progressed face is composited into the ground truth photo to match the hairstyle and background (see supplementary material for comparisons of just the face regions).

Morphs

Video on YouTube

Left image is the starting input photo, and right image will transform to age 80 to show our automatic aging process.
Age 1 Age 2 Age 3 Age 3

Dataset

The FGNet Aging Database that we used for comparisons: FGNET.zip The dataset is by Andreas Lanitis, Cyprus College. Htmls of our Amazon Mechanical Turk tasks used for evaluation: data.

Acknowledgements

We thank Google and Intel for supporting this research and "Thunder" for allowing us to use his photo collection for comparisons.

Popular press:

The Seattle Times | UW news | CSE news | Ars Technica | GeekWire | Gizmodo | Gizmag | Telegraph UK | IEEE Spectrum | PetaPixel | DailyMail UK | Independent UK | Popular Science | חורים ברשת | Wired UK | Ynet | NBC Today | Shape
TV and Radio interviews:
NBC's Today Show | KOMO News | KUOW/NPR | KING5 | KIRO TV

Visit NBCNews.com for breaking news, world news, and news about the economy

BibTex

Contact

For more information please contact: Ira Kemelmacher-Shlizerman

Last modified: