Real time motion estimation using a neural architecture implemented on GPUs

This work describes a neural network based architecture that represents and estimates object motion in videos. This architecture addresses multiple computer vision tasks such as image segmentation, object representation or characterization, motion analysis and tracking. The use of a neural network architecture allows for the simultaneous estimation of global and local motion and the representation of deformable objects. This architecture also avoids the problem of finding corresponding features while tracking moving objects. Due to the parallel nature of neural networks, the architecture has been implemented on GPUs that allows the system to meet a set of requirements such as: time constraints management, robustness, high processing speed and re-configurability. Experiments are presented that demonstrate the validity of our architecture to solve problems of mobile agents tracking and motion analysis.     

A backstepping controller for path‐tracking of an underactuated autonomous airship

In this paper we propose a nonlinear control approach for the path‐tracking of an autonomous underactuated airship. A backstepping controller is designed from the airship nonlinear dynamic model including wind disturbances, and further enhanced to consider actuators saturation. Control implementation issues related to airship underactuation are also addressed, namely control allocation and an attitude reference shaping to obtain a faster error correction with smoother input requests. The results obtained demonstrate the capacity of an underactuated unmanned airship to execute a realistic mission including vertical take‐off and landing, stabilization and path‐tracking, in the presence of wind disturbances, with a single robust control law. Copyright © 2008 John Wiley & Sons, Ltd.     

Formalizing interoperability for test case generation purpose

This study deals with interoperability testing of protocol implementations. The objective is to define a method for automatic test derivation based on formal definitions. First, the notion of interoperability criteria is introduced. They formally describes the conditions that two implementations must verify in order to be considered interoperable. Then, based on the equivalence of two of the defined interoperability criteria, we propose a method to derive automatically interoperability test cases that avoids the state-space explosion problem.     

Nanocellulose-Carboxymethylcellulose Electrolyte for Stable,High-Rate Zinc-Ion Batteries

Aqueous Zn ion batteries (ZIBs) are one of the most promising battery chemistries for grid-scale renewable energy storage. However, their application is limited by issues such as Zn dendrite formation and undesirable side reactions that can occur in the presence of excess free water molecules and ions. In this study, a nanocellulose-carboxymethylcellulose (CMC) hydrogel electrolyte is demonstrated that features stable cycling performance and high Zn²⁺ conductivity (26 mS cm⁻¹), which is attributed to the material's strong mechanical strength (≈70 MPa) and water-bonding ability. With this electrolyte, the Zn-metal anode shows exceptional cycling stability at an ultra-high rate, with the ability to sustain a current density as high as 80 mA cm⁻² for more than 3500 cycles and a cumulative capacity of 17.6 Ah cm⁻² (40 mA cm⁻²). Additionally, side reactions, such as hydrogen evolution and surface passivation, are substantially reduced due to the strong water-bonding capacity of the CMC. Full Zn||MnO₂ batteries fabricated with this electrolyte demonstrate excellent high-rate performance and long-term cycling stability (>500 cycles at 8C). These results suggest the cellulose-CMC electrolyte as a promising low-cost, easy-to-fabricate, and sustainable aqueous-based electrolyte for ZIBs with excellent electrochemical performance that can help pave the way toward grid-scale energy storage for renewable energy sources.     

Imposing temporal consistency on deep monocular body shape and pose estimation

Computational Visual Media - Accurate and temporally consistent modeling of human bodies is essential for a wide range of applications, including character animation, understanding human social...     

Synergistic photoluminescent interaction of Si and CdTe quantum dots

Microsystem Technologies - We report the synergistic photoluminescent effect observed in heterogeneous colloidal solutions comprising different volumetric concentrations of Si and CdTe quantum dots...     

Engineered Multifunctional Albumin‐Decorated Porous Silicon Nanoparticles for FcRn Translocation of Insulin

The last decade has seen remarkable advances in the development of drug delivery systems as alternative to parenteral injection‐based delivery of insulin. Neonatal Fc receptor (FcRn)‐mediated transcytosis has been recently proposed as a strategy to increase the transport of drugs across the intestinal epithelium. FcRn‐targeted nanoparticles (NPs) could hijack the FcRn transcytotic pathway and cross the epithelial cell layer. In this study, a novel nanoparticulate system for insulin delivery based on porous silicon NPs is proposed. After surface conjugation with albumin and loading with insulin, the NPs are encapsulated into a pH‐responsive polymeric particle by nanoprecipitation. The developed NP formulation shows controlled size and homogeneous size distribution. Transmission electron microscopy (TEM) images show successful encapsulation of the NPs into pH‐sensitive polymeric particles. No insulin release is detected at acidic conditions, but a controlled release profile is observed at intestinal pH. Toxicity studies show high compatibility of the NPs with intestinal cells. In vitro insulin permeation across the intestinal epithelium shows approximately fivefold increase when insulin is loaded into FcRn‐targeted NPs. Overall, these FcRn‐targeted NPs offer a toolbox in the development of targeted therapies for oral delivery of insulin.     

Metrics for MT evaluation: evaluating reordering

Translating between dissimilar languages requires an account of the use of divergent word orders when expressing the same semantic content. Reordering poses a serious problem for statistical machine translation systems and has generated a considerable body of research aimed at meeting its challenges. Direct evaluation of reordering requires automatic metrics that explicitly measure the quality of word order choices in translations. Current metrics, such as BLEU, only evaluate reordering indirectly. We analyse the ability of current metrics to capture reordering performance. We then introduce permutation distance metrics as a direct method for measuring word order similarity between translations and reference sentences. By correlating all metrics with a novel method for eliciting human judgements of reordering quality, we show that current metrics are largely influenced by lexical choice, and that they are not able to distinguish between different reordering scenarios. Also, we show that permutation distance metrics correlate very well with human judgements, and are impervious to lexical differences.