Robust person re-identification via graph convolution networks

Person re-identification (re-id) aims to identity the same person over multiple cameras; it has been successfully applied to various computer vision applications as a fundamental method. Owing to the development of deep learning, person re-id methods, which typically use triplet networks based on triplet loss, have demonstrated great success. However, the appearances of people are similar and hence difficult to distinguish in many cases. Therefore, we present a novel graph convolution network and enhances traditional triplet loss functions. Our method defines reference, positive, and negative features for triplet loss as three vertices of a graph, respectively, and adjusts their mutual distance through learning. The method adopts graph convolutions efficiently, thereby affording low computational costs. Experimental results demonstrate that our method is superior to the baseline on the Market-1501 dataset. The proposed GCN-based triplet loss considerably contributes to improve re-identification methods quantitatively and qualitatively.

     

Effect of asymmetrical channel chirping on field intensity patterns of supermodes and modal discrimination in index guided laser arrays

The effect of channel-width chirping on near- and far-field intensity patterns of the six supermodes was investigated. The supermode discrimination was evaluated in various channel-chirped index guided laser arrays. The results show that the linearly channel-chirped laser array has very good supermode discrimination which is better than that of a uniform laser array, the V channel-chirped laser array has the smallest radiation angle of the fundamental supermode among the calculated arrays, and the asymmetrically V channel-chirped array has a very small radiation angle of the fundamental supermode, which is smaller than that of the uniform array and also allows for very good fundamental supermode discrimination against the higher-order supermodes, which is better than that of the V channel-chirped laser array.     

A computational approach to characterize formation of a passivation layer in lithium metal anodes

Li metal anode is the “Holy Grail” material of advanced Lithium-ion-batteries (LIBs). However, it is plagued by uncontrollable dendrite growth resulting in poor cycling efficiency and short-circuiting of batteries. This has spurred a plethora of research to understand the underlying mechanism of dendrite formation. While experimental studies suggest that there are complex physical and chemical interactions between heterogeneous solid-electrolyte interphase (SEI) and dendrite growth, most of the studies do not reveal the mechanisms triggering these interactions. To deal with this knowledge gap, we propose a multiscale modeling framework which couples kinetic Monte Carlo and Molecular Dynamics simulations. Specifically, the model has been developed to account for (a) heterogeneous SEI, (b) dendrite-SEI interactions, and (c) effect of electrolyte on Li electrodeposition and potential dendrite formation. This allows the proposed computational model to be extended to various electrolytes and SEI species and generate results consistent with previous experimental studies.     

Chemiresistor sensor using elastomer-functionalized carbon nanotube nanocomposites for the detection of gasoline spills

Two types of multi-walled carbon nanotube (MWNT)-based elastomer nanocomposites are used as a sensor material for the detection of gasoline spills by applying the interdigitated electrode (IDE) device. MWNT-g-polyisoprene (PI) and Si-MWNT/natural rubber (NR) are prepared by applying “grafting-from” and “grafting-to” process, respectively. When compared based on the identical condition of gasoline sensing test, the maximum response value to the exposure of gasoline is 17.5 for MWNT-g-PI sensor and 12.9 for Si-MWNT/NR sensor, which reach the maximum in less than 3 min. The MWNT-g-PI sensor selectively detects gasoline, and its response is completely reversible. It shows that the longer chain length of PI brings about the larger response of MWNT-g-PI sensor to gasoline. The sensitivity of MWNT-g-PI sensor highly depends on both how much gasoline is exposed to the sensor and what bias voltage is applied to the IDE device. The IDE sensor using MWNT-g-PI nanocomposites effectively detects gasoline spills.     

Temperature and grain size dependence of superplasticity in a Zn−0.3wt.%Al alloy

The superplastic deformation behavior of quasi-single phase Zn-0.3 wt. %Al was investigated. A series of load relaxation and tensile tests was conducted at various temperatures ranging from RT (20 °C) to 200 °C. The recently proposed internal variable theory of structural superplasticity was applied. The flow curves obtained from load relaxation tests were shown to consist of contributions from interface sliding (IS) and accommodating plastic deformation. In the case of quasi-single phase Zn-0.3 wt.% Al alloy with an average agrain size of 1 μm, the IS behavior could be described as a viscous flow process characterized by a power index of M_g=0.5. A large elongation of about 1400% was obtained at room temperature and the strain rate sensitivity parameter was about 0.4. Although relatively large-grained (10 μm) single phase alloy showed a high value of strain rate sensitivity comparable to that of fine-grained alloy at very low strain rate range, IS was not expected from the analysis based on the internal variable theory of structural superplasticity at room temperature. As the temperature increased above 100 °C, however, the contribution from IS was observed at a very low strain rate range. A high elongation of ∼400% was obtained in a specimen of 10-μm-grain-size at 200 °C under a strain rate of 2×10⁻⁴/sec. Jointly appointed at Center for Advanced Aerospace Materials (CAAM)     

Effect of yttrium on the stability of aluminide-yttrium composite coatings in a cyclic high-temperature hot-corrosion environment

A composite coating of aluminide-yttrium has shown excellent corrosion resistance in a cyclic high-temperature hot-corrosion environment. To understand the effect of yttrium on the stability of the composite coating, the specimens were prepared with various coating parameters of Y thickness, sequence of post heat treatment and surface condition before Y-ion plating. Performance of the composite coating was evaluated by isothermal oxidation and cyclic high-temperature hot corrosion. Isothermal-oxidation-test results show that the Y in the composite coating helps to form a thick and dense Al₂O₃ scale which is ductile and resistant to thermal stress. The Y in Al₂O₃ may act as a donor which leads to an increase in concentration of interstitial oxygen and, thus, increases in oxidation rate. The presence of Y₂O₃ and (Y, Al) O-type compounds in grain boundaries of Al₂O₃ and boundaries between the Al₂O₃ and NiAl effectively prohibits the fast diffusion of oxidants (such as O and S) and Al along grain boundaries. Consequently, it may induce slow diffusion through the matrix, and thus the corrosion resistance of the composite coating under cyclic hot corrosion increases substantially.     

Analysis of deformation texture inhomogeneity and stability condition of shear components in f.c.c. metals

The texture of cold rolled aluminum sheet has been known to vary through the thickness due to inhomogeneous deformation during rolling. The copper texture is obtained in the center layer that is plane strain compressed while the shear texture in the surface layer, which is approximated by major 001 011 and minor 111 112 and 111 110 components. The stability condition of these components was calculated based on the full constraint Taylor-Bishop-Hill theory and could be described by a parameterdɛ₁₃/dɛ₁₁ with the suffixes 1 and 3 indicating the rolling and thickness directions, respectively.     

Preparation of epitaxial and polycrystalline bismuth titanate thin films on single crystal (100) MgO by chemical solution deposition

Epitaxial and polycrystalline Bi₄Ti₃O₁₂ thin films were prepared on single crystal (100) MgO substrates by a chemical solution deposition process using metal naphthenates as starting materials. Pyrolyzed films (at 500°C) were annealed for 30 min in air at 650, 700, 750 and 800°C, respectively. The effects of annealing temperature on the crystallinity, epitaxy and surface morphology of the films were investigated by X-ray diffraction θ-2θ scans, pole-figure analysis, and atomic force microscopy (AFM). Epitaxially grown films annealed at 700 and 750°C, respectively, showed growth of three-dimensional needle-shaped grains. During annealing at 800°C, grain growth of Bi₄Ti₃O₁₂ may be suppressed by the formation of a titanium-rich phase such as Bi₂Ti₂O₇ owing to Bi volatilization, resulting in lower root mean square roughness than that of film annealed at 750°C.     

Boron distribution in a low-alloy steel

Boron distribution in a low-alloy steel (15B26:0.25C-0.29Cr-0.03Ti-0.028Al-0.0016B) has been characterized employing Fission Track Etching (FTE) method. The characteristics of boron distribution with variation of cooling rate after austenitization and through case-hardened depth after carburization were analyzed. Hardenability of 15B26 steel was also evaluated through Jominy-end-quench test and the results are as follows: It was observed that, in austenitized 15B26 steel, boron was distributed uniformly over the whole area of specimen with a little segregation along the austenite grain boundaries at higher cooling rates and boron precipitates were formed in the intergranular as well as transgranular regions at lower cooling rates. Jominy equivalents (HRC 35) of 15B26 steel were fairly increased between the Jominy temperatures of 820°C and 850°C, which might result from the increase of the amount of soluble boron in austenite due to the dissolution of borocarbides between 820°C and 850°C. In carburized 15B26 steel, the different through thickness features of boron distribution from the carburized surface were found; coarse nodular boron precipitates up to the depth of 150 μm; uniform distribution of dissolved boron between 150~650 μm; and segregation of boron atoms along grain boundaries in the regions deeper than 650 μm.     

A timing controller embedded driver IC with 3.24‐Gbps eDP interface for chip‐on‐glass TFT‐LCD applications

This paper presents a timing controller embedded driver (TED) IC with 3.24‐Gbps embedded display port (eDP), which is implemented using a 45‐nm high‐voltage CMOS process for the chip‐on‐glass (COG) TFT‐LCD applications. The proposed TED‐IC employs the input offset calibration scheme, the zero‐adjustable equalizer, and the phase locked loop‐based bang‐bang clock and data recovery to enhance the maximum data rate. Also, the proposed TED‐IC provides efficient power management by supporting advanced link power management feature of eDP standard v1.4. Additionally, the smart charge sharing is proposed to reduce the dynamic power consumption of output buffers. Measured result demonstrates the maximum data rate of 3.24 Gbps from a 1.1 V supply voltage with a 7.9‐inch QXGA 60‐Hz COG‐LCD prototype panel and 44% power saving from the display system.