An ideal approach to the problem of pose-invariant face recognition would handle continuous pose variations, would not be database specific, and would achieve high accuracy without any manual intervention. Most of the existing approaches fail to match one or more of these goals. In this paper, we present a fully automatic system for pose-invariant face recognition that not only meets these requirements but also outperforms other comparable methods. We propose a 3D pose normalization method that is completely automatic and leverages the accurate 2D facial feature points found by the system. The current system can handle 3D pose variation up to +-45 in yaw and +-30 in pitch angles. Recognition experiments were conducted on the USF 3D, Multi-PIE, CMU-PIE, FERET, and FacePix databases. Our system not only shows excellent generalization by achieving high accuracy on all 5 databases but also outperforms other methods convincingly. International Conference on Computer Vision (ICCV)This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonprofit educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved. AbstractAn ideal approach to the problem of pose-invariant face recognition would handle continuous pose variations, would not be database specific, and would achieve high accuracy without any manual intervention. Most of the existing approaches fail to match one or more of these goals. In this paper, we present a fully automatic system for poseinvariant face recognition that not only meets these requirements but also outperforms other comparable methods. We propose a 3D pose normalization method that is completely automatic and leverages the accurate 2D facial feature points found by the system. The current system can handle 3D pose variation up to ±45• in yaw and ±30• in pitch angles. Recognition experiments were conducted on the USF 3D, Multi-PIE, CMU-PIE, FERET, and FacePix databases. Our system not only shows excellent generalization by achieving high accuracy on all 5 databases but also outperforms other methods convincingly.

We present an unsupervised learning approach to recover 3D human pose from 2D skeletal joints extracted from a single image. Our method does not require any multiview image data, 3D skeletons, correspondences between 2D-3D points, or use previously learned 3D priors during training. A lifting network accepts 2D landmarks as inputs and generates a corresponding 3D skeleton estimate. During training, the recovered 3D skeleton is reprojected on random camera viewpoints to generate new 'synthetic' 2D poses. By lifting the synthetic 2D poses back to 3D and re-projecting them in the original camera view, we can define self-consistency loss both in 3D and in 2D. The training can thus be self supervised by exploiting the geometric selfconsistency of the lift-reproject-lift process. We show that self-consistency alone is not sufficient to generate realistic skeletons, however adding a 2D pose discriminator enables the lifter to output valid 3D poses. Additionally, to learn from 2D poses 'in the wild', we train an unsupervised 2D domain adapter network to allow for an expansion of 2D data. This improves results and demonstrates the usefulness of 2D pose data for unsupervised 3D lifting. Results on Human3.6M dataset for 3D human pose estimation demonstrate that our approach improves upon the previous unsupervised methods by 30% and outperforms many weakly supervised approaches that explicitly use 3D data.

3D pose estimation from a single image is a challenging task in computer vision. We present a weakly supervised approach to estimate 3D pose points, given only 2D pose landmarks. Our method does not require correspondences between 2D and 3D points to build explicit 3D priors. We utilize an adversarial framework to impose a prior on the 3D structure, learned solely from their random 2D projections. Given a set of 2D pose landmarks, the generator network hypothesizes their depths to obtain a 3D skeleton. We propose a novel Random Projection layer, which randomly projects the generated 3D skeleton and sends the resulting 2D pose to the discriminator. The discriminator improves by discriminating between the generated poses and pose samples from a real distribution of 2D poses. Training does not require correspondence between the 2D inputs to either the generator or the discriminator. We apply our approach to the task of 3D human pose estimation. Results on Human3.6M dataset demonstrates that our approach outperforms many previous supervised and weakly supervised approaches.

We theoretically study the optical tomography of maximally entangled states generated at the output modes of a beam splitter. We consider even and odd coherent states in one of the input modes and vacuum state in the other input mode of the beam splitter. We have shown

15Landfast sea ice is an important seasonal feature along most Arctic coastlines, 16 such as that of the Chukchi Sea near Barrow, Alaska. Its stability throughout the ice 17 season is determined by many factors but grounded pressure ridges are the primary 18 stabilizing component. Landfast ice breakouts occur when these grounded ridges 19 fail or unground, and previously stationary ice detaches from the coast and drifts 20 away. Using ground-based radar imagery from a coastal ice and ocean observatory 21at Barrow, we have developed a method to estimate the extent of grounded ridges 22 by tracking ice motion and deformation over the course of winter and have derived 23

We theoretically study the optical tomography of the time evolved states generated by the evolution of different kinds of initial wave packets in a Kerr medium. Exact analytical expression for the optical tomogram of the quantum state at any instant during the evolution of a generic initial wave packet is derived in terms of Hermite polynomials. Time evolution of the optical tomogram is discussed for three kinds of initial states: a coherent state, an mphoton-added coherent state, and even and odd coherent states. We show the manifestation of revival and fractional revivals in the optical tomograms of the time evolved states. We find that the optical tomogram of the time evolved state at the instants of fractional revivals shows structures with sinusoidal strands. The number of sinusoidal strands in the optical tomogram of the time evolved state at l-sub-packet fractional revivals is l times the number of sinusoidal strands present in the optical tomogram of the initial state. We have also investigated the effect of decoherence on the optical tomograms of the states at the instants of fractional revivals for the initial states considered above. We consider amplitude decay and phase damping models of decoherence, and show the direct manifestations of decoherence in the optical tomogram.

Taking a look at both sides of the ice: comparison of ice thickness and drift speed as 1 observed from moored, airborne and shore--based instruments near Barrow, Alaska 2 3 Andrew R.

Environmental change and increasing industrial activity in the maritime Arctic require strategies to adapt to change and ensure safe operations. This problem has been defined at the broader strategic level. We evaluate key aspects of environmental security in ice-covered waters, focusing on tactical and operational information needs, which have received less attention. Monitoring of environmental hazards and effective emergency response in sea ice environments require high-resolution data of ice hazard distributions (e.g., multiyear ice, landfast ice breakout, and ice push events), ice movement and deformation, as well as ice characteristics and dynamics relevant to emergency response. We have developed a prototype coastal observing system at Barrow, Alaska, that addresses such information needs. Imagery obtained from a marine X-band radar with a digital controller is combined with data from on-ice sensors (ice thickness, ice and water temperature, sea level) and assessments of potentially hazardous ice conditions by local experts. Digital imagery and data are processed and disseminated in near-real time. Using a combination of image processing approaches (optical flow, Lucas-Kanade tracker), ice velocity fields, floe trajectories, and boundaries of stationary ice are derived automatically. Early onset of hazardous events is detected through Hidden Markov Modeling, providing potential decision support in operational settings. We evaluate the utility of the system and strategies towards integration with broader emergency response efforts.