Person re-identification based on multi-level feature complementarity of cross-attention with part metric learning

Person re-identification is an image retrieval task, and its task is to perform a person matching in different cameras by a given person target. This research has been noticed and studied by more and more people. However, pose changes and occlusions often occur during a person walking. Especially in the most related methods, local features are not used to simply and effectively solve the problems of occlusion and pose changes. Moreover, the metric loss functions only consider the image-level case, and it cannot adjust the distance between local features well. To tackle the above problems, a novel person re-identification scheme is proposed. Through experiments, we found that we paid more attention to different parts of a person when we look at him from a horizontal or vertical perspective respectively. First, in order to solve the problem of occlusion and pose changes, we propose a Cross Attention Module (CAM). It enables the network to generate a cross attention map and improve the accuracy of person re-identification via the enhancement of the most significant local features of persons. The horizontal and vertical attention vectors of the feature maps are extracted and a cross attention map is generated, and the local key features are enhanced by this attention map. Second, in order to solve the problem of the lack of expression ability of the single-level feature maps, we propose a Multi-Level Feature Complementation Module (MLFCM). In this module, the missing information of high-level features is complemented by low-level features via short skip. Feature selection is also performed among deep features maps. The purpose of this module is to get the feature maps with complete information. Further, this module solves the problem of missing contour features in high-level semantic features. Third, in order to solve the problem that the current metric loss function cannot adjust the distance between local features, we propose Part Triple Loss Function (PTLF). It can reduce both within-class and increase between-class distance of the person parts. Experimental results show that our model achieves high values on Rank-k and mAP on Market-1501, Duke-MTMC and CUHK03-NP.

     

Robust image hashing using progressive feature selection for tampering detection

The main problem addressed in this paper is the robust tamper detection of the image received in a transmission under various content-preserving attacks. To this aim the progressive feature point selection method is proposed to extract the feature points of high robustness; with which, the local feature and color feature are then generated for each feature point. Afterwards, the robust image hashing construction method is proposed by using the location-context information of the features. The constructed hash is attached to the image before transmission and it can be used for analyzing at destination to filter out the geometric transformations occurred in the received image. After image restoration, the similarity of the global hashes between the source image and restored image is calculated to determine whether the received image has the same contents as the trusted one or has been maliciously tampered. When the received image being judged as a tampered image, the hashes calculated with the proposed Horizontal Location-Context Hashing (HLCH) and Vertical Location-Context Hashing (VLCH) methods will be used to locate the tampered regions. Experimental results on different images with tampering of arbitrary size and location demonstrate that our image authentication and tampering localization scheme are superior to the state-of-the-art methods under various attacks.     

Vapor transport equilibration-induced formation of micron-sized ErNbO4 precipitates in Zn/Er-codoped LiNbO3 crystals

Congruent LiNbO₃ crystals with Zn/Er-doping levels of 5.5/1, 6/0.15 and 7/0.8 mol%/mol% were thermally treated at 1100 °C over 101 h using vapor transport equilibration (VTE) technique. It is shown that the VTE treatment has brought these crystals closer to stoichiometric composition. X-ray powder diffraction and Er³⁺ spectroscopic characterizations show that the VTE treatment induced formation of micron-sized ErNbO₄ precipitates on the surface and in the bulk of all crystals studied. It was observed using an optical microscope that the precipitates grow preferably along X, Y and Z axes of the host matrix, and show different morphologies, densities and dimensions, depending on the Er³⁺ and Zn³⁺ doping level. The formation origin of the precipitate is qualitatively explained.     

Embedding Pyramids into 3D Meshes

The pyramid architecture is a powerful topology in the area of computer vision. On the other hand, the 3D mesh architecture possesses rich topological features which make it suitable for building scalable parallel processor systems. The usefulness of these two architectures has led us to consider the problem of embedding pyramids into 3D meshes, for which we present two solutions. The first solution, termednatural embedding,maps a pyramid into a 3D mesh such that each level of the pyramid is mapped to a single level of the 3D mesh. The second solution, termedmultiple embedding,allows simultaneous embedding of multiple pyramids into a single 3D mesh. The quality of both solutions is evaluated using dilation and expansion measures. Using the multiple embedding, we are able to obtain an average dilation of 1.26 and a near-optimal expansion of 1.12.     

Modeling secondary organic aerosol formation via multiphase partitioning with molecular data

A new model for atmospheric secondary organic aerosol (SOA) is presented for biogenic compounds. It is based to the extent possible on experimental molecular SOA data, and it is compatible with any existing gas-phase chemical kinetic mechanism. Six SOA precursors or groups of precursors are used to represent biogenic monoterpenes and sesquiterpenes. SOA formation is modeled using five SOA surrogates to represent classes of compounds with different partitioning properties, e.g., hydrophobicity, aqueous solubility, acid dissociation, and saturation vapor pressure. Model simulations are evaluated against smog chamber data for SOA yields and some adjustments are made to uncertain stoichiometric coefficients and saturation vapor pressure parameters to improve model performance. The model is applied undertypical atmospheric conditions to exemplify the effect of relative humidity on SOA formation and the relative contributions of hydrophilic and hydrophobic SOA.     

Evaluation of the Dynamic Yield Locus Tester (DYLT)

ABSTRACT

Shear cells based on the constant volume approach are a promising alternative to the conventional constant load techniques. The objectives of this research were to evaluate the dynamic yield locus tester (DYLT) based on the constant volume approach, and the computer controlled Jenkie tester (CCJT) based on the constant load approach. This was done by comparing the flow parameters obtained using the two techniques for BCR limestone, glass fibers, ground silica, microcrystalline cellulose, and wheat flour. Test results showed the flow parameters obtained using the DYLT and the CCJT were comparable at a level of significance of 0.05 for BCR limestone, glass fibers, ground silica, and microcrystalline cellulose. There were significant differences (p > 0.05) between the flow parameters of the DYLT and CCJT for wheat flour.     

Sensitivity of particulate matter nitrate formation to precursor emissions in the California San Joaquin Valley

The formation of secondary ammonium nitrate during the 1995 Integrated Monitoring Study (IMS95) in San Joaquin Valley, CA was investigated using a box model that simulates the atmospheric chemistry and gas/particle partition of inorganic compounds. The concentration of particulate matter (PM) nitrate was found to be sensitive to reductions in VOC emissions. Nitric acid, rather than ammonia, was the limiting reagent in the formation of PM nitrate. The formation of nitric acid was more sensitive to the availability of oxidants than that of NOx. Oxidant chemistry in wintertime conditions in the San Joaquin Valley was shown to be VOC-sensitive. In fact, a decrease in NOx emissions may have the counter-intuitive effect of increasing PM nitrate.     

Sensitized luminescence and energy transfer in Ce and Eu codoped calcium magnesium chlorosilicate

Ce³⁺ and Eu²⁺ ions codoped calcium magnesium chlorosilicate [Ca₈Mg(SiO₄)₄Cl₂] phosphors have been synthesized and characterized. Intense bluish-green light was observed under ultraviolet and blue light excitations. The diffuse reflection, excitation and emission spectra of Ca₈Mg(SiO₄)₄Cl₂:Ce³⁺,Eu²⁺ were measured at room temperature, and enhancement of Eu²⁺ emission was confirmed. The fluorescence lifetimes of Ce³⁺ were measured, and the mechanism of nonradiative energy transfer from Ce³⁺ to Eu²⁺ is found to be electric dipole–dipole interaction. The probabilities and efficiencies of nonradiative energy transfer were also calculated, and the results indicate that the efficiency can be as high as 86% in this material system.     

Diffusion Properties of Tm3+ in Congruent LiNbO3 Crystal

Diffusion properties of Tm³⁺ in congruent LiNbO₃ crystal have been investigated, together with other two related issues, i.e., Tm³⁺‐doping contribution to refractive index of LiNbO₃ substrate and Li out‐diffusion. Four X‐cut and four Z‐cut congruent LiNbO₃ substrates locally coated with 15–31 nm‐thick Tm‐metal films were annealed in surrounding air under different temperatures of 1030°C–1130°C for different durations of 20–70 h. After anneal, refractive index at Tm³⁺‐doped and Tm³⁺‐free parts of crystal surface was measured at the wavelengths of 1311 and 1553 nm and surface Li₂O contents were evaluated from measured refractive index. The results show that Tm³⁺ doping has a weak effect on substrate index and a small contribution to index increment in waveguide layer in comparison with Ti⁴⁺‐ or Zn²⁺ doping. The Li₂O content at the Tm³⁺‐doped surface equals that at the Tm³⁺‐free surface. The Li out‐diffusion depends mainly on the diffusion temperature. Below 1100°C, the Li out‐diffusion is not measurable. At 1130°C, a 30‐h diffusion procedure may cause 0.2–0.3 mol% slight loss of Li₂O content. Secondary ion mass spectrometry was used to study the Tm³⁺ diffusion properties. The results show that the diffused Tm³⁺ ions in all samples follow a complementary error function profile. From measured Tm³⁺ profiles, characteristic diffusion parameters such as diffusivity, diffusion constant, activation energy, solubility, solubility constant, and heat of solution were obtained and discussed in comparison with the case of Er³⁺ diffusion. In comparison with Er³⁺ diffusion, the Tm³⁺ diffusion shows similar anisotropy and temperature dependence of solubility. In the aspect of diffusivity, under lower temperature the Tm³⁺ has a lower diffusivity than the Er³⁺, and their diffusivity difference reduces with the increased temperature and becomes null at 1130°C.     

Comparative studies of spectroscopic properties in Er–Yb codoped phosphate glasses

A spectroscopic investigation of an extensive series of commercially available Er³⁺–Yb³⁺ codoped phosphate glasses (IOG-1) with different Er–Yb concentrations have been presented based upon spectroscopic measurements and Judd–Ofelt theory. The composition and the atomic concentration of the respective elements of the glass samples were analyzed. Various spectroscopic parameters were obtained to evaluate the performances of these glasses as a laser material in the eye-safe laser wavelength of 1.53 μm. Using the measured fluorescence lifetimes the radiative quantum efficiency of the ⁴I_13/2 → ⁴I_15/2 transition is estimated to be 100% for the glass substrates with an Er³⁺ concentration of lower than 2.0 × 10²⁰ ions/cm³. The stimulated emission and absorption cross-sections were also determined and compared. The infrared and upconversion fluorescence were studied, and under 975 nm excitation the dominant upconversion mechanisms are excited state absorption (ESA) and energy transfer (ET) for the green emission, and ET for the red emission. The data obtained here provide useful guidelines on the choice of IOG-1 glasses with the fixed Er–Yb concentrations.