•We propose a DCNN with multi-tasks learning for adult image classification.•Our network seamlessly integrates the global and local contexts.•It achieves excellent performance in three-class adult ...image classification.
With the wide availability of the Internet and the proliferation of pornographic images online, adult image detection and filtering has become very important to prevent young people from reaching these harmful contents. However, due to the large diversity in adult images, automatic adult image detection is a difficult task. In this paper, a new deep convolutional neural network (DCNN) based approach is proposed to classify images into three classes, i.e. porn, sexy, and benign. Our approach takes both the entire picture (global context) and the meaningful region (local context) information into consideration. The proposed network is composed of three parts, i.e. the image characteristics subnet to extract discriminative low-level image features, the sensitive body part detection subnet to detect adult-image related regions, and the feature extraction and fusion subnet to generate high-level features for image classification. A multi-task learning scheme is designed to optimize the network with both the global and local information. Experiments are carried out on two datasets with over 160,000 images. From the experiment results, it was observed that the proposed network achieved high classification accuracies (96.6% in the AIC dataset and 92.7% in the NPDI dataset) and outperformed the other approaches investigated.
Following the identification and confirmation of the substructures of the leading-edge vortex (LEV) system on flapping wings, it is apparent that the actual LEV structures could be more complex than ...had been estimated in previous investigations. In this experimental study, we reveal for the first time the detailed three-dimensional (3-D) flow structures and evolution of the LEVs on a flapping wing in the hovering condition at high Reynolds number (Re=1624). This was accomplished by utilizing an electromechanical model dragonfly wing flapping in a water tank (mid-stroke angle of attack=60 degrees) and applying phase-lock based multi-slice digital stereoscopic particle image velocimetry (DSPIV) to measure the target flow fields at three typical stroke phases: at 0.125 T (T=stroke period), when the wing was accelerating; at 0.25 T, when the wing had maximum speed; and at 0.375 T, when the wing was decelerating. The result shows that the LEV system is a collection of four vortical elements: one primary vortex and three minor vortices, instead of a single conical or tube-like vortex as reported or hypothesized in previous studies. These vortical elements are highly time-dependent in structure and show distinct ;stay properties' at different spanwise sections. The spanwise flows are also time-dependent, not only in the velocity magnitude but also in direction.
Pancreatic ductal adenocarcinoma (PDAC) is lethal, and the majority of patients present with locally advanced or metastatic disease that is not amenable to cure. Thus, with surgical resection being ...the only curative modality, it is critical that disease is identified at an earlier stage to allow the appropriate therapy to be applied. Unfortunately, a specific biomarker for early diagnosis has not yet been identified; hence, no screening process exists. Recently, high-throughput screening and next-generation sequencing (NGS) have led to the identification of novel biomarkers for many disease processes, and work has commenced in PDAC. Genomic data generated by NGS not only have the potential to assist clinicians in early diagnosis and screening, especially in high-risk populations, but also may eventually allow the development of personalized treatment programs with targeted therapies, given the large number of gene mutations seen in PDAC. This review introduces the basic concepts of NGS and provides a comprehensive review of the current understanding of genetics in PDAC as related to discoveries made using NGS.
A novel CuAlMn/polymer/carbon nanomaterials multi-scale composite with damping capability was successfully prepared. These composite materials composed of a CuAlMn memory alloy foam skeleton and a ...polymer matrix, which was in turn reinforced with different carbon nanomaterials. Experiment results show that the loss factor of CuAlMn/polymer/CNT composite is above 0.1, while the loss factor of CuAlMn/polymer/(Au@CNT + CNF + GNS) composite is higher than 0.12 over the entire temperature range, with the highest value reaching about 0.15. Also, it can be seen that the storage modulus of CuAlMn/polymer composite is above 3500 MPa, while CuAlMn/polymer/(Au@CNT + CNF + GNS) composite is about 1000 MPa. The loss factor for CuAlMn/polymer/(Au@CNT + CNF + GNS) composite is highest, and the storage modulus of it is lowest. A three-phase model was introduced to analyze the damping mechanism of the composite. The results indicate that the damping performance of the prepared composites is mainly affected by interfacial loss, which can be effectively improved by increasing the mismatch of elastic modulus near the interface and/or increasing the interfacial area. On these bases, the optimal performance of the composite material was predicted by calculation and simulation. This multi-scale composite offers a tremendously promising future, which needs further thorough exploration.
Display omitted
Lithium metal anode with high theoretical capacity has attracted lots of attention, but dendrite growth results in poor cyclic stability during the process of repeated plating/stripping. Herein, a ...scaffold constructed by surface growth of hierarchical lithiophilic acicular CoNiO2 on the porous Ni foam (NF@CoNiO2) is synthesized and used to construct three-dimensional (3D) host for lithium metal anode. The experimental results show that the lithium dendrite growth is effectively inhibited and satisfactory cycling performance is achieved. Benefiting from the stable lithiophilic sites, good conductivity from the NF, large porosity, and high strength 3D structure, the host enable Li||NF@CoNiO2@Li cells to deliver low hysteresis voltage (7.7 mV) and ultra-long cycling stability (more than 2800 h, 1400 cycles) at 1 mA cm−2 with a capacity of 1 mAh cm−2. The full cells configured the NF@CoNiO2@Li anode with LiFePO4 still display a reversible discharge capacity of 126.2 mAh g−1 after 140 cycles with a Coulombic efficiency of 92.5% at 1C. This work paves a feasible way to construct the lithiophilic skeleton for dendrite-free lithium metal anode.
Display omitted
•Face-centered-cubic (FCC) Cr, L12 Cr3Cu, C14 Cr2Nb, C15 Cr2Nb, and (Cr, Cu)2Nb were detected in the alloy. A time–temperature dependent phase transition map was established.•The ...competition of nucleation and growth of L12 Cr3Cu and C14 Cr2Nb during aged at 600 °C was discussed.•It is discussed why C15 Cr2Nb is precipitated during the solidification process of the alloy, but C14 Cr2Nb is precipitated during the aging process.
Cu-Cr-Nb alloys have attracted many researchers’ interest for decades while their structure and second phases inside were unclear. We have characterized the second phase particles formed during solidification, hot rolling, and aging treatments in a Cu-4.06Cr-1.25Nb alloy. Face-centered-cubic (FCC) Cr, L12 Cr3Cu, C14 Cr2Nb, C15 Cr2Nb, and Nb-rich particles were detected in the alloy. The crystal structure, precipitation, phase transformation behavior and orientation relationship (OR) were uncovered. C15 Cr2Nb particles formed in the solidification process while C14 Cr2Nb particles precipitated as aged at 600 °C. The competition of nucleation between nano-scale Cr3Cu and Cr2Nb particles aged at 600 °C was discussed. The nucleation of L12 Cr3Cu was prior to that of C14 Cr2Nb, and the precipitation of Cr2Nb particles would limit the coarsening of Cr particles. A phase transformation map during age treatment was established on the basis of transmission electron microscope (TEM) images and X-ray diffraction (XRD) patterns. These findings are helpful in developing high-performance Cu-Cr-Nb alloys.
Display omitted
•A novel pea-pod-like TiO2@MoSe2/RGO composite was built.•A secondary substrate was constructed and the synergistic effect was used.•The strong interfacial coupling was realized ...through chemical bonds formation.•A high specific capacity of ∼446 mAh/g with CE ∼90 % was achieved.
Sodium-ion hybrid capacitors (SIHCs) have received extensive attention and research due to their combined advantages of high power performance and high energy density. However, the lack of high-performance anode materials has been a challenge for sodium storage devices. Herein, a heteroatom co-doped hollow spherical TiO2@MoSe2/reduced graphene oxide (RGO) with pea pods structure is designed and synthesized. This unique composite is made of TiO2 as the stable framework, MoSe2 sheet as the main active materials, and RGO conductive network as the protective cover. The sandwich structure provides stable and ordered transport channels for electrons and Na+, and greatly increases the number of active sites. As a result, TiO2@MoSe2/RGO exhibits satisfactory electrochemical performance. At a high current density of 2 A g-1, TiO2@MoSe2/RGO still maintains a capacity of 106 mAh g-1 after 1000 cycles. The hybrid device possesses an energy density of 135 Wh kg−1 at power density of 1667 W kg−1, demonstrating the design concept of heterogeneous composite provides a new route for the development of electrode materials in sodium-ion storage devices.
Genome-wide single nucleotide polymorphism (SNP) association studies recently identified four SNPs (rs10757274, rs2383206, rs2383207, and rs10757278) on chromosome 9p21 that were associated with ...coronary artery disease (CAD) and myocardial infarction (MI) in Caucasian populations from northern Europe and North America. Our aim was to determine whether these SNPs were associated with MI in a southern Europe/Mediterranean population. We employed a case-control association design involving 416 MI patients and 308 non-MI controls from Italy. Significant allelic association was identified between all four SNPs and MI. The association remained significant after adjusting for covariates for MI (P=0.007-0.029). One risk haplotype (GGGG; P=0.028) and one protective haplotype (AAAA; P=0.047) were identified. Genotypic association analysis demonstrated that the SNPs conferred susceptibility to MI most likely in a dominant model (P=0.0007-0.013). When the case cohort was divided into a group of MI patients with a family history (n=248) and one group without it (n=168), the positive, significant association was identified only in the group with the family history. These results indicate that chromosome 9p21 confers risk for development of MI in an Italian population.