In this paper we provide, to the best of our knowledge, the first Bayesian formulation of one of the most successful and well-studied statistical models of shape and texture, i.e. Active Appearance Models (AAMs). To this end, we use a simple probabilistic model for texture generation assuming both Gaussian noise and a Gaussian prior over a latent texture space. We retrieve the shape parameters by formulating a novel cost function obtained by marginalizing out the latent texture space. This results in a fast implementation when compared to other simultaneous algorithms for fitting AAMs, mainly due to the removal of the calculation of texture parameters. We demonstrate that, contrary to what is believed regarding the performance of AAMs in generic fitting scenarios, optimization of the proposed cost function (using models learned from small amounts of training data) produces results comparable to those obtained with discriminatively trained (using several thousand training images) state-of-the-art methods in the problem of facial alignment “in the wild”.
Tracking results
Results obtained with a model trained using only ~800 training images.
Code
- Stay tuned, a link to the code will be uploaded soon.
Publications
- J. Alabort-i-Medina, S. Zafeiriou, Bayesian Active Appearance Models, CVPR 2014. [pdf] [poster]