Search

Since scene graph can be used as the basis of many high-level vision semantic tasks, scene graph generation has attracted more and more attention of researchers, but most works are limited by the long-tail distribution of dataset, tend to predict those frequent but uninformative predicates such as "on" and "of." From a novel perspective, we found that in the training process, the model would promote the categories included in the batch, while suppressing the categories not in the batch. The long-tailed distribution of the data leads to the continuous suppression of tail categories, thus results in the model bias. In order to solve the problem above, we propose a novel simple and effective method named Batch Equalization Loss, which can be applied to most of the existing models and can bring effective improvement with only a few changes. It is worth noting that our method can achieve a more significant improvement on small batches than on big batches. Extensive experiments on the VG150 dataset show that our work can bring significant improvement on the basis of existing works. Code will be available at GitHub in the near future. [ABSTRACT FROM AUTHOR]

Extracting planes from a three-dimensional (3D) point cloud is a challenging problem for many applications with 3D point clouds. In this paper, a novel fast plane extraction method based on the point pair feature (PPF) is proposed. There are two stages included in the proposed method. One is the local processing stage to sample some points in a point cloud and calculate their PPF descriptors. In this stage, the coplanar property of the PPF is used to extract initial planes from the sampling points. The other one is a global processing stage to consider all the other points in the point cloud, and assess whether they are located in the initial planes by calculating the distance from each point to the initial planes. We can extract and determine the final planes in the global processing stage. Compared with the efficient random sample consensus (RANSAC) and the 3D kernel-based Hough transform (3DKHT), the results show that for the complex scene, the extracting time of our method is less than 0.3% of the RANSAC method, and the precision rate of our method is about 9% and 17% higher than that of the RANSAC method and 3DKHT method, respectively. [ABSTRACT FROM AUTHOR]

A facile method was developed in this work for the batch production of fibrous microelectrodes in enzyme-free electrochemical glucose sensors. The microelectrodes utilized carbon fibers (CFs) as the substrate and cobalt-containing metal organic framework, ZIF-67, as the electrocatalyst, which were mediated with reduced graphene oxide (rGO). The overall fabrication processes were carried out with carbon clothes and the fibrous ZIF-67/rGO/CF microelectrodes were extracted from the resulting composite, rendering a time-saving and economical batch-production approach. As the working electrode (WE) for electrochemical glucose detection, ZIF-67/rGO/CF manifested a wide linear detection range of 1–1000 μM, a low detection limit of 0.2 μM (S/N = 3), accompanied with the excellent selectivity and stability, which were comparable to the recently reported enzyme-free electrodes based on cobalt. A fibrous electrode based glucose sensor with ZIF-67/rGO/CF as the WE was further assembled, which showed reliable detection capability towards the glucose in serum and demonstrated the intriguing potential of ZIF-67/rGO/CF in practical applications. [ABSTRACT FROM AUTHOR]

The point pair feature (PPF) algorithm is one of the best-performing 3D object recognition algorithms. However, the high dimensionality of its search space is a disadvantage of this algorithm. This high dimensionality means the feature matching process contains a large number of uninformative features, which reduces recognition speed. To solve this problem and improve the object recognition speed, this paper proposes a fast 3D object recognition algorithm based on the plane-constrained point pair features. By utilizing the property of the coplanar point pair features and the characteristics of the object placement plane, the proposed algorithm extracts the object placement plane through convex hull area calculation, eliminates irrelevant point pair features, and then performs object recognition with the reduced point pair feature descriptors for the feature matching. Experimental results demonstrate that the proposed algorithm significantly reduces the number of feature descriptors and accelerates the recognition speed of 3D objects in a complex background. Compared to the original point pair feature algorithm, the proposed method can achieve better performance and efficiency for 3D object recognition. [ABSTRACT FROM AUTHOR]

All-solid-state potassium ion-selective electrodes (K+ ISEs) using ordered mesoporous carbon (OMC) sphere as the solid contact (SC) are fabricated in this work, which exhibit excellent K+-sensing capability including the stable Nernstian response in a wide linear range of 10−4.19–10−0.21 M, a low detection limit of 5.4 μM, short response time, and good stability. The excellent performances of the K+ ISEs are attributable to the OMC spheres with interconnected mesopore structure, uniform spherical morphology, high surface area, high conductivity, and high capacitance, which provide the highly efficient ion-to-electron transduction between the ion-selective membrane and conductive component in the electrodes. More importantly, the OMC sphere-based SC layer can be deposited in the flexible ISEs by solution processing, which thus enables the construction of ion-sensitive field-effect transistor-based handheld K+-sensing system with high sensitivity and reproducibility, demonstrating the potentials of the OMC spheres in practical ion detection devices. [ABSTRACT FROM AUTHOR]

Sodium-ion batteries (SIBs) is considered as a promising alternative to lithium-ion batteries. Supercapacitors (SCs) are receiving great attention for their significantly higher power density than batteries and prolonged cycle life. Herein, SIBs and SCs based on N-doped amorphous multi-size pores dominated polymeric frameworks were fabricated and examined. The enlarged interlayer spacing and multi-size-pore dominated interconnected architecture with high specific surface area, high pore volume and high N content optimize the electrochemical performance of N-PPF-20. As an anode material, N-PPF-20 exhibited a sodium ion storage capacity of 432.2 mAh g−1 at a current density of 0.05 A g−1, while maintaining a reversible capacity of 61.1 mAh g−1 at an ultrahigh current density of 20 A g−1. Additionally, a specific capacity of 158.3 mAh g−1 at 1 A g−1 was obtained after 1000 cycles, indicating an excellent cycling stability. When tested as an electrode material for SCs, N-PPF-20 delivered a high specific capacitance of 438.7 F g−1 at 0.1 A g−1, and a specific capacitance of 111.2 F g−1 was achieved even at a high current density of 10 A g−1. Meanwhile, a long-term cycling life test demonstrated a specific capacitance of 120 F g−1 at an ultrahigh current density of 10 A g−1 after 10,000 cycles. [ABSTRACT FROM AUTHOR]

A facile and cost-efficient synthesis approach toward hierarchically porous cobalt aluminum layered double hydroxide with hydrangea flower-like morphology (PCALs) is developed in this work via the hydrothermal treatment of cobalt- and aluminum-containing precursors under basic conditions following etching with NaOH. The architectures and porosities of the obtained PCALs can be easily modified by the fabrication conditions such as the concentration of the starting materials and the time of etching, which can further influence their electrochemical behavior. As the electrode material in a three-electrode supercapacitor, the PCALs can deliver a high capacitance of 550 F g for 1000 cycles with a retention rate over 82% at 10 A g. In addition, the asymmetric capacitor using PCALs as cathode and active carbon as anode also exhibits an excellent capacitance of 103.5 F g at 0.5 A g, indicating their potentials in the applications of energy storages. [ABSTRACT FROM AUTHOR]

Three-dimensional hybrids of cobalt oxide (CoO) and graphene frameworks are fabricated via a facile hydrothermal self-assembly process. By adjusting the time of the hydrothermal treatment, the morphologies of the CoO components can be modified from rods to nanoparticles, which further manifest influences on the electrochemical performance of the hybrids. As the anode in lithium-ion battery, the hybrid loaded with spherical CoO nanoparticles exhibits the highest reversible capacity of 1148 mA h g at 100 mA g for 100 cycles among the three samples. Even at a high current density of 5000 mA g, its reversible capacity is still kept at 600 mA h g, outperforming the reported hybrids of CoO and graphene. [ABSTRACT FROM AUTHOR]

Construction of organic semiconducting materials with in-plane π-conjugated structures and robustness through carbon-carbon bond linkages, alternatively as organic graphene analogs, is extremely desired for powerfully optoelectrical conversion. However, the poor reversibility for sp2 carbon bond forming reactions makes them unavailable for building high crystalline well-defined organic structures through a self-healing process, such as covalent organic frameworks (COFs). Here we report a scalable solution-processing approach to synthesize a family of two-dimensional (2D) COFs with trans-disubstituted C = C linkages via condensation reaction at arylmethyl carbon atoms on the basis of 3,5-dicyano-2,4,6-trimethylpyridine and linear/trigonal aldehyde (i.e., 4,4″-diformyl-p-terphenyl, 4,4′-diformyl-1,1′-biphenyl, or 1,3,5-tris(4-formylphenyl)benzene) monomers. Such sp2 carbon-jointed-pyridinyl frameworks, featuring crystalline honeycomb-like structures with high surface areas, enable driving two half-reactions of water splitting separately under visible light irradiation, comparable to graphitic carbon nitride (g-C3N4) derivatives. Semiconducting covalent organic frameworks (COF) with conjugated structures are desired for optoelectric conversion but poor reversibility of bond forming reactions hampers the formation of crystalline structures. Here, the authors synthesise COFs with disubstituted C = C linkages which enable water splitting under visible light irradiation. [ABSTRACT FROM AUTHOR]

In this work, a novel soft-hard template method towards the direct fabrication of graphene films on silicon/silica substrate is developed via a tri-constituent self-assembly route. Using cetyl trimethyl ammonium bromide (CTAB) as a soft template, silica (SiO2) from tetramethoxysilane as a hard template, and pyrene as a carbon source, the self-assembly process allows the formation of a sandwich-like SiO2/CTAB/pyrene composite, which can be further converted to high quantity graphene films with a thickness of ~1 nm and a size of over 5 μm by thermal treatment. The morphology and thickness of the graphene films can be effectively controlled through the adjustment of the ratio of pyrene to CTAB. Furthermore, a high nonlinear refractive index n2 of ~10−12 m2 W−1 is measured from graphene/silica hybrid film, which is six orders of magnitude larger than that of silicon and comparable to the graphene from chemical vapor deposition process. [ABSTRACT FROM AUTHOR]