Face recognition from video

In spite of over two decades of intense research, illumination and pose invariance remain prohibitively challenging aspects of face recognition for most practical applications. The objective of this work is to recognize faces using video sequences both for training and recognition input, in a realistic, unconstrained setup in which lighting, pose and user motion pattern have a wide variability and face images are of low resolution. In particular there are three areas of novelty: (i) we show how a photometric model of image formation can be combined with a statistical model of generic face appearance variation, learnt offline, to generalize in the presence of extreme illumination changes; (ii) we use the smoothness of geodesically local appearance manifold structure and a robust same-identity likelihood to achieve invariance to unseen head poses; and (iii) we introduce an accurate video sequence "reillumination" algorithm to achieve robustness to face motion patterns in video. We describe a fully automatic recognition system based on the proposed method and an extensive evaluation on 171 individuals and over 1300 video sequences with extreme illumination, pose and head motion variation. On this challenging data set our system consistently demonstrated a nearly perfect recognition rate (over 99.7%), significantly outperforming state-of-the-art commercial software and methods from the literature. © Springer-Verlag Berlin Heidelberg 2006.

An ARGONAUTE4-containing nuclear processing center colocalized with Cajal bodies in Arabidopsis thaliana.

ARGONAUTE4 (AGO4) and RNA polymerase IV (Pol IV) are required for DNA methylation guided by 24 nucleotide small interfering RNAs (siRNAs) in Arabidopsis thaliana. Here we show that AGO4 localizes to nucleolus-associated bodies along with the Pol IV subunit NRPD1b; the small nuclear RNA (snRNA) binding protein SmD3; and two markers of Cajal bodies, trimethylguanosine-capped snRNAs and the U2 snRNA binding protein U2B''. AGO4 interacts with the C-terminal domain of NRPD1b, and AGO4 protein stability depends on upstream factors that synthesize siRNAs. AGO4 is also found, along with the DNA methyltransferase DRM2, throughout the nucleus at presumed DNA methylation target sites. Cajal bodies are conserved sites for the maturation of ribonucleoprotein complexes. Our results suggest a function for Cajal bodies as a center for the assembly of an AGO4/NRPD1b/siRNA complex, facilitating its function in RNA-directed gene silencing at target loci.

RNAi, DRD1, and histone methylation actively target developmentally important Non-CG DNA methylation in Arabidopsis

Cytosine DNA methylation protects eukaryotic genomes by silencing transposons and harmful DNAs, but also regulates gene expression during normal development. Loss of CG methylation in the Arabidopsis thaliana met1 and ddm1 mutants causes varied and stochastic developmental defects that are often inherited independently of the original met1 or ddm1 mutation. Loss of non-CG methylation in plants with combined mutations in the DRM and CMT3 genes also causes a suite of developmental defects. We show here that the pleiotropic developmental defects of drm1 drm2 cmt3 triple mutant plants are fully recessive, and unlike phenotypes caused by met1 and ddm1, are not inherited independently of the drm and cmt3 mutations. Developmental phenotypes are also reversed when drm1 drm2 cmt3 plants are transformed with DRM2 or CMT3, implying that non-CG DNA methylation is efficiently re-established by sequence-specific signals. We provide evidence that these signals include RNA silencing though the 24-nucleotide short interfering RNA (siRNA) pathway as well as histone H3K9 methylation, both of which converge on the putative chromatin-remodeling protein DRD1. These signals act in at least three partially intersecting pathways that control the locus-specific patterning of non-CG methylation by the DRM2 and CMT3 methyltransferases. Our results suggest that non-CG DNA methylation that is inherited via a network of persistent targeting signals has been co-opted to regulate developmentally important genes. © 2006 Chan et al.