MFF-Net: Multimodal Feature Fusion Network for 3D Object Detection

In complex traffic environment scenarios, it is very important for autonomous vehicles to accurately perceive the dynamic information of other vehicles around the vehicle in advance. The accuracy of 3D object detection will be affected by problems such as illumination changes, object occlusion, and object detection distance. To this purpose, we face these challenges by proposing a multimodal feature fusion network for 3D object detection (MFF-Net). In this research, this paper first uses the spatial transformation projection algorithm to map the image features into the feature space, so that the image features are in the same spatial dimension when fused with the point cloud features. Then, feature channel weighting is performed using an adaptive expression augmentation fusion network to enhance important network features, suppress useless features, and increase the directionality of the network to features. Finally, this paper increases the probability of false detection and missed detection in the non-maximum suppression algorithm by increasing the one-dimensional threshold. So far, this paper has constructed a complete 3D target detection network based on multimodal feature fusion. The experimental results show that the proposed achieves an average accuracy of 82.60% on the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) dataset, outperforming previous state-of-the-art multimodal fusion networks. In Easy, Moderate, and hard evaluation indicators, the accuracy rate of this paper reaches 90.96%, 81.46%, and 75.39%. This shows that the MFF-Net network has good performance in 3D object detection.     

Image downscaling via co-occurrence learning

Image downscaling is one of the widely used operations in image processing and computer graphics. It was recently demonstrated in the literature that kernel-based convolutional filters could be modified to develop efficient image downscaling algorithms. In this work, we present a new downscaling technique which is based on kernel-based image filtering concept. We propose to use pairwise co-occurrence similarity of the pixelpairs as the range kernel similarity in the filtering operation. The co-occurrence of the pixel-pair is learned directly from the input image. This co-occurrence learning is performed in a neighborhood based fashion all over the image. The proposed method can preserve the high-frequency structures, which were present in the input image, into the downscaled image. The idea is further extended to the case of fractions factor of downscaling. The resulting images retain visually-important details and do not suffer from edge-blurring artifact. We demonstrate the effectiveness of our proposed approach with extensive experiments on a large number of images downscaled with various downscaling factors.     

Insight into the wetting and interfacial behavior of Cu-Ti/ZrB2 system: A combined experimental and DFT calculation

The wetting of ZrB₂ ceramic by molten pure Cu, Cu-(20, 30, 40, 50 at.%)Ti alloys were performed by the sessile drop method under vacuum of ∼4 × 10⁻⁴ Pa in the 1030−1070 °C temperature range. The wetting and interfacial behaviors of Cu-Ti/ZrB₂ systems were analyzed and discussed. Furthermore, the work of adhesion and electronic properties of the Cu(1 1 1)/ZrB₂(0 0 0 1) and Ti(0 0 0 1)/ZrB₂(0 0 0 1) atomic interfaces were quantitatively evaluated by first-principles method. The wettability is significantly improved with the Ti concentration or wetting temperature increasing as a result of the formation of (Ti) solid solution at interface. The calculated results show that the B-terminated ZrB₂ (0 0 0 1) surface is preferable to bond with Ti(0 0 0 1) surface, and the electronic structure reveals that the dominant interfacial bonding is the Ti-B ionic bond for the Ti/B-terminated ZrB₂ interface.     

Digital light processing of Si-based composite ceramics and bulk silica ceramics from a high solid loading polysiloxane/SiO2 slurry

Nowadays, 3D polymer-derived silicon oxycarbide ceramics (SiOC) can be fabricated by the stereolithography method successfully. However, due to their intrinsically poor ceramic content and the large thermal shrinkage during the pyrolysis, it is difficult for the bulk 3D polysiloxane precursors to be pyrolyzed into dense 3D-ceramics. Herein, the ceramic content of the photocurable polysiloxane precursors was increased by adding a large amount of SiO₂ powders into the low solid content epoxy-acrylic siloxane. After being added with proper dispersant, the viscosity of the high solid loading polysiloxane/SiO₂ slurry can reach a proper level. Bulk 3D-silica ceramics with the wall-thickness around 8 mm can be conveniently fabricated from the polysiloxane/SiO₂ slurry by stereolithography and a two-step sintering process. The addition of a proper sintering aid NaF can promote the 3D-PSO/SiO₂ precursor to be converted into dense and crack-free 3D-silica ceramics with good mechanical proprieties.     

Surface Corrections for Remote Sensing Reflectance in Case 2 Waters of Lake Superior

We establish the average wavelength dependence for rough-surface reflectance S_rs(λ) in Lake Superior and determine its magnitude at any individual site by extrapolating the total reflectance or raw remote sensing reflectance R_rsT(λ) measured at the surface to the asymptotic limit of zero scattering where S_rs(λ) ≅ R_rsT(λ). The results show that S_rs(λ) differs from the flat-surface Fresnel reflectance of sky radiance used in standard determination of the remote sensing reflectance R_rs(λ) attributed to the scattering of light by particles and molecules in bulk water. In waters containing colored dissolved organic matter (CDOM), R_rs(λ) can be very low and radiometric measurements at ground level can often lead to negative estimates of R_rs(λ) if we assume simple mirror-like reflectance of the sky radiance as the basis for correcting radiometric data for surface reflectance. We examine the differences between Fresnel reflectance and S_rs(λ) and estimate the ratio of the concentrations of CDOM versus suspended particles that could produce negative values of R_rs(λ) if assume Fresnel reflectance in calculation of R_rs(λ).     

On the penetration energy for fibre-reinforced plastics under low-velocity impact conditions

This paper deals with the prediction of the penetration energy for fibre-reinforced plastics subjected to low-velocity impact. Some results available in the literature, allowing evaluation of the main parameters affecting the energy-absorbing capacity of a composite laminate, are reviewed first. It is shown that, for a given fibre type, the penetration energy is substantially influenced by the total fibre volume and tup diameter, whereas other factors, such as resin type and content, fibre architecture, stacking sequence and orientations, play a secondary role in the phenomenon. An empirical power law equation recently proposed by the authors, from which the penetration energy can be evaluated, is then assessed on the basis of experimental data previously published. The results indicate that the exponent of the power law is probably independent of the material considered, being practically the same for carbon- and glass-fibre-reinforced plastics, and even for an isotropic material such as polycarbonate, prone to extensive plastic yielding before final failure. The formula proposed, useful for in-plane isotropic and moderately anisotropic composites, can also permit the comparison of impact data generated under different impact conditions.     

高效求解三维装箱问题的剩余空间最优化算法

为实现三维装箱问题的高效求解，提出了一个三维的剩余空间最优化算法（Three-Dimensional Residual-Space-Optimized Algorithm，3D-RSO）。在满足3个著名约束的条件下，该算法将三维问题转化为带有高度约束的二维问题，通过对箱子放置后的剩余空间状态分析，提出了基于概率较优的空间分割方法和箱子布置规则。相比于传统算法，3D-RSO在求解过程中不需要任何的预处理和搜索操作，是一种最坏计算复杂度为[O(2n2)]的直接求解算法。针对强异构体的实验表明，该算法能够在极短的时间内对算例进行高效求解，适合应用在大规模或者需要被快速求解的三维装箱问题中。     

焊接专业课教学的点滴心得

文章从教学实践出发,分析了压焊方法与设备课程的传统教学方法,提出了相应的教学改革新思路,较好地解决了传统教学跟不上新技术发展需要的矛盾.     

Microstructure,charge carrier conduction mechanism model,dielectric properties and leakage current analysis of Dy2FeMnO6 nanomaterial

The structure, microstructure, charge transport properties, dielectric properties, and leakage current density of Dy₂FeMnO₆ synthesized by the auto combustion method are investigated in this study. The monoclinic crystal structure of the sample with space group P2₁/n was observed. The CoO₆ and MnO₆ octahedra were found distorted and tilted. The calculation of the bond valence sum confirmed the +3 and + 3/+4 oxidation states of Fe and Mn, respectively. The average crystallite, particle, and grain sizes were 43.01, 94, and 187.68 nm, respectively. The anticipated stoichiometry of the constituent elements was confirmed by energy–dispersive X–ray spectra. The frequency–dependent conductivity obeyed Jonscher’s power law, and the frequency exponent associated with this law increased with temperature, indicating the non–overlapping small polaron tunnelling (NSPT) mechanism for charge conduction. The enhancement of the dielectric loss tangent with temperature was consistent with the dc conductivity. The frequency dependence of dielectric properties was explained using Havriliak–Negami formalism. The sample followed Vogel–Fulcher law, which is similar to the relaxor ferroelectrics, and the ferroelectricity increased with frequency. Conduction, migration, and dielectric relaxation all have similar activation energies. The sample exhibited low leakage current density, which followed Schottky emission with a barrier height of 0.02(4) eV. The obtained results were compared with other rare–earth–based double perovskites.     

NEXAFS spectroscopy of CVD diamond films exposed to fusion relevant hydrogen plasma

A series of CVD diamond films have been exposed to hydrogen plasma in the linear magnetized plasma device, MAGPIE, with various applied sample stage biases between 0 V (no applied bias) to − 500 V. The plasma-induced damage to the surface structure of the diamond films has been investigated by Near Edge X-ray Absorption Fine Structure (NEXAFS) spectroscopy in both the Auger electron yield (AEY) and total fluorescence yield (TFY) modes. The key diamond NEXAFS spectral features (diamond core exciton and second absolute band gap) are found to be diminished following plasma exposure as measured in the surface sensitive, AEY spectra, whilst these features remain unchanged relative to an unexposed diamond reference film as measured using the bulk sensitive, TFY spectra. These results, in conjunction with SRIM simulations, show definitively that the damage to the surface of the diamond films is restricted to the scale of the penetration depth of the H ions and no damage is induced at greater depths. The power and sensitivity of NEXAFS spectroscopy in assessing damage to the surface of diamond from fusion-relevant plasma-surface interactions are demonstrated.