We propose a new method to quickly and accurately predict 3D positions of body joints from a single depth image, using no temporal information. We take an object recognition approach, designing an intermediate body parts representation that maps the difficult pose estimation problem into a simpler per-pixel classification problem. Our large and highly varied training dataset allows the classifier to estimate body parts invariant to pose, body shape, clothing, etc. Finally we generate confidence-scored 3D proposals of several body joints by reprojecting the classification result and finding local modes.The system runs at 200 frames per second on consumer hardware. Our evaluation shows high accuracy on both synthetic and real test sets, and investigates the effect of several training parameters. We achieve state of the art accuracy in our comparison with related work and demonstrate improved generalization over exact whole-skeleton nearest neighbor matching.

No abstract

This is a book about the problem of vision. How is it that a torrent of data from a television camera, or from biological visual receptors, can be reduced to perceptions-the recognition of familiar objects and the concise description of unfamiliar ones? There is of course an immense literature in psychophysics 1 , neurophysiology and neuroanatomy that provides some answers in the case of biological systems (see Uttal (1981) for a taxonomy). For instance, the functioning of light-sensitive cells in mammalian vision is understood in some detail (Marks et al. 1964); and the elegant, orderly, spatial correspondence of feature detectors in the brain with the array of cells in the retina, is well known (Hubel and Wiesel 1968). There has also been much dialogue between psychophysics and neurophysiology/neuroanatomy. Examples are the discovery of spatial bandpass channels (Campbell and Robson 1968, Braddick et al. 1978), and understanding the perception of coloured light (Livingstone and Hubel 1984, Jameson and Hurvich 1961) and surface colour (Land 1983, Zeki 1983). These instances are but parts of a very large body of knowledge of biological vision. Over the last two decades, computers have introduced a new strand into the study of vision. The earliest work (Roberts, 1965) produced systems able to recognise simple objects and manipulate them in a controlled way (Ambler et al. 1975). These systems were, of course, vastly inferior to the biological systems studied by the psychophysicists, neuroanatomists 1 Psychophysics is the application of physical methods to the study of psychological properties. Visual psychophysics typically probes the mechanisms of human vision by noting a subject's perception of specially designed patterns, under controlled experimental conditions.

No abstract

No abstract

The main challenge in articulated body motion tracking is the large number of degrees of freedom (around 30)

No abstract

We report the results of a search for νe appearance in a νµ beam in the MINOS long-baseline neutrino experiment. With an improved analysis and an increased exposure of 8.2 × 10 20 protons on the NuMI target at Fermilab, we find that 2 sin 2 (θ23) sin 2 (2θ13) < 0.12 (0.20) at 90% confidence level for δ=0 and the normal (inverted) neutrino mass hierarchy, with a best fit of 2 sin 2 (θ23) sin 2 (2θ13) = 0.041−0.031 (0.079−0.053 ). The θ13=0 hypothesis is disfavored by the MINOS data at the 89% confidence level.PACS numbers: 14.60. Pq, 14.60.Lm, arXiv:1108.0015v1 [hep-ex] 29 Jul 2011 2 It has been experimentally established that neutrinos undergo flavor change as they propagate [1][2][3][4][5][6][7]. This phenomenon is well-described by three-flavor neutrino oscillations, characterized by the spectrum of neutrino masses together with the elements of the PMNS mixing matrix [8]. This matrix is often parametrized by three Euler angles θ ij and a CP-violating phase δ. While θ 12 and θ 23 are known to be large [1,4,6], θ 13 appears to be relatively small [9][10][11][12][13], with the tightest limits so far coming from the CHOOZ [10] and MINOS [12] experiments. The T2K collaboration has recently reported indications of a nonzero value for θ 13 at the 2.5σ confidence level (C.L.) [14]. This letter reports new θ 13 constraints from the MINOS experiment, using an increased data set and significant improvements to the analysis.MINOS is a two-detector long-baseline neutrino oscillation experiment situated along the NuMI neutrino beamline [15]. The 0.98-kton Near Detector (ND) is located on-site at Fermilab, 1.04 km downstream of the NuMI target. The 5.4-kton Far Detector (FD) is located 735 km downstream in the Soudan Underground Laboratory. The two detectors have nearly identical designs, each consisting of alternating layers of steel (2.54 cm thick) and plastic scintillator (1 cm). The scintillator layers are constructed from optically isolated, 4.1 cm wide strips that serve as the active elements of the detectors. The strips are read out via optical fibers and multi-anode photomultiplier tubes. Details can be found in Ref. [16].The data used in this analysis come from an exposure of 8.2×1020 protons on the NuMI target. The corresponding neutrino events in the ND have an energy spectrum that peaks at 3 GeV and a flavor composition of 91.7% ν µ , 7.0% ν µ , and 1.3% ν e +ν e , as estimated by beamline and detector Monte Carlo (MC) simulations, with additional constraints from MINOS ND data and external measurements [6,17]. The two-detector arrangement and the relatively small intrinsic ν e component make this analysis rather insensitive to beam uncertainties. Neutrino-nucleus and final-state interactions are simulated using NEUGEN3 [18], and particle propagation and detector response are simulated with GEANT3 [19].MINOS is sensitive to θ 13 through ν µ → ν e oscillations. To leading order, the probability for this oscillation mode is given bywhere ∆m 2 32 (in units of eV 2 ) and θ 23 are the dominant atmospheric oscillation...