糖心TV

Abstract: Many face recognition algorithms use ``distance-based'' methods: feature
vectors are extracted from each face and distances in feature space are
compared to determine matches. In this paper we argue for a fundamentally
different approach. We consider each image as having been generated from an
underlying cause (a latent identity variable, or LIV).  In recognition we
evaluate the probability that two faces have the same underlying cause.
Since image generation is noisy, we can never be exactly certain what this
cause was, so we integrate (marginalize) over all possible causes. We
present examples of identification and verification and show that the LIV
approach outperforms equivalent distance-based algorithms. Moreover, other
advantages include: (i) a natural approach to changes in pose and lighting
(ii) the ability to implement novel algorithms that have no distance-based
equivalent (iii) a principled way to combine multiple observations and prior
information.  Finally we demonstrate how this framework can be applied to
the more general problem of clustering face images.