Global Contextual Dependency Network for Object Detection

Current two-stage object detectors extract the local visual features of Regions of Interest (RoIs) for object recognition and bounding-box regression. However, only using local visual features will lose global contextual dependencies, which are helpful to recognize objects with featureless appearances and restrain false detections. To tackle the problem, a simple framework, named Global Contextual Dependency Network (GCDN), is presented to enhance the classification ability of two-stage detectors. Our GCDN mainly consists of two components, Context Representation Module (CRM) and Context Dependency Module (CDM). Specifically, a CRM is proposed to construct multi-scale context representations. With CRM, contextual information can be fully explored at different scales. Moreover, the CDM is designed to capture global contextual dependencies. Our GCDN includes multiple CDMs. Each CDM utilizes local Region of Interest (RoI) features and single-scale context representation to generate single-scale contextual RoI features via the attention mechanism. Finally, the contextual RoI features generated by parallel CDMs independently are combined with the original RoI features to help classification. Experiments on MS-COCO 2017 benchmark dataset show that our approach brings continuous improvements for two-stage detectors.

PointSCNet: Point Cloud Structure and Correlation Learning Based on Space-Filling Curve-Guided Sampling

Geometrical structures and the internal local region relationship, such as symmetry, regular array, junction, etc., are essential for understanding a 3D shape. This paper proposes a point cloud feature extraction network named PointSCNet, to capture the geometrical structure information and local region correlation information of a point cloud. The PointSCNet consists of three main modules: the space-filling curve-guided sampling module, the information fusion module, and the channel-spatial attention module. The space-filling curve-guided sampling module uses Z-order curve coding to sample points that contain geometrical correlation. The information fusion module uses a correlation tensor and a set of skip connections to fuse the structure and correlation information. The channel-spatial attention module enhances the representation of key points and crucial feature channels to refine the network. The proposed PointSCNet is evaluated on shape classification and part segmentation tasks. The experimental results demonstrate that the PointSCNet outperforms or is on par with state-of-the-art methods by learning the structure and correlation of point clouds effectively.

The Power of Many: A Physarum Swarm Steiner Tree Algorithm

This paper presents a novel explore-and-fuse approach to solving a large array of problems that cannot be solved by traditional divide-and-conquer. This approach is inspired by Physarum, a unicellular slime mold capable of solving the traveling salesman and Steiner tree problems. Besides exhibiting individual intelligence, Physarum can also share information with other Physarum organisms through fusion. Inspired by the characteristics of Physarum, we spawn many Physarum organisms to explore the problem space in parallel, each gathering information and forming partial solutions pertaining to a local region of the problem space. When the organisms meet, they fuse and share information, eventually forming one organism which has a global view of the problem and can apply its intelligence to find an overall solution to the problem. We demonstrate this novel approach on the NP-hard Steiner tree problem, developing the Physarum Steiner Algorithm. This algorithm is of particular interest due to its ability to leverage parallel processing, avoid obstacles, and operate on various shapes and topological surfaces including the rectilinear grid.

Neighborhood co-occurrence modeling in 3D point cloud segmentation

A significant performance boost has been achieved in point cloud semantic segmentation by utilization of the encoder-decoder architecture and novel convolution operations for point clouds. However, co-occurrence relationships within a local region which can directly influence segmentation results are usually ignored by current works. In this paper, we propose a neighborhood co-occurrence matrix (NCM) to model local co-occurrence relationships in a point cloud. We generate target NCM and prediction NCM from semantic labels and a prediction map respectively. Then, Kullback-Leibler (KL) divergence is used to maximize the similarity between the target and prediction NCMs to learn the co-occurrence relationship. Moreover, for large scenes where the NCMs for a sampled point cloud and the whole scene differ greatly, we introduce a reverse form of KL divergence which can better handle the difference to supervise the prediction NCMs. We integrate our method into an existing backbone and conduct comprehensive experiments on three datasets: Semantic3D for outdoor space segmentation, and S3DIS and ScanNet v2 for indoor scene segmentation. Results indicate that our method can significantly improve upon the backbone and outperform many leading competitors.

Approach for Obtaining Material Mechanical Properties in Local Region of Structure Based on Accurate Analysis of Micro-indentation Test

The hot or cold processing would induce the change and the inhomogeneous of the material mechanical properties in the local processing region of the structure, and it is difficult to obtain the specific mechanical properties in these regions by using the traditional material tensile test. To accurately get actual material mechanical properties in the local region of structure, a micro-indentation test system incorporated by an electronic universal material test device has been established. An indenter displacement sensor and a group of special micro-indenter assemblies are established. A numerical indentation inversion analysis method by using ABAQUS software is also proposed in this study. Based on the above test system and analysis platform, an approach to obtaining material mechanical properties in the local region of structures is proposed and established. The ball indentation test is performed and combined with the energy method by using various changed mechanical properties of 316L austenitic stainless steel under different elongations. The investigated results indicate that the material mechanical properties and the micro-indentation morphological changes have evidently relevance. Compared with the tensile test results, the deviations of material mechanical parameters, such as hardness H, the hardening exponent n, the yield strength σy, and others are within 5% obtained through the indentation test and the finite element analysis. It provides an effective and convenient method for obtaining the actual material mechanical properties in the local processing region of the structure.

Perception of e-Learning Among Hungarian Engineering Students

The success of developing e-learning is determined by curriculum quality, the availability of technological requirements and, to an even greater extent, by user response to the technology introduced in the process. Digital enhancement of education through e-learning solutions should also take the unique attributes of the targeted local region, institution, target group and field of expertise into consideration prior to implementation. The paper reports on research conducted to understand the approach of engineering students (n=94) to e-learning supported by Moodle in Hungary, including computer use, evaluation of e-learning materials, systems and online exams. The research used an online questionnaire for exploring the motivation and restrictive factors of using e-learning. Survey findings confirm that e-learning functions primarily in a complementary way to traditional learning. 87.2% of the students in the sample just download the learning materials and use those offline often or regularly. 58% of them find the usefulness of e-learning materials good, but structure or aesthetics is evaluated weak or moderate by more than half of the respondents. Considering the exams, 38% of the students with previous experience in online exams prefer the traditional exams, while 25% prefer the online format. Since access to technological tools and services required for effective e-learning is available, continuous training of the teachers and tutors is necessary both for developing their everyday skills and recognizing the LCMS opportunities.

The Mie representation for Mercury’s magnetospheric currents

Poloidal–toroidal magnetic field decomposition is a useful application of the Mie representation and the decomposition method enables us to determine the current density observationally and unambiguously in the local region of magnetic field measurement. The application and the limits of the decomposition method are tested against the Mercury magnetic field simulation in view of BepiColombo’s arrival at Mercury in 2025. The simulated magnetic field data are evaluated along the planned Mercury Planetary Orbiter (MPO) trajectories and the current system that is crossed by the spacecraft is extracted from the magnetic field measurements. Afterwards, the resulting currents are classified in terms of the established current system in the vicinity of Mercury. Graphical Abstract

Analysis on Drug-Resistance-Associated Mutations among Multidrug-Resistant Mycobacterium tuberculosis Isolates in China

As the causative bacteria of tuberculosis, Mycobacteriumtuberculosis (M. tb) is aggravated by the emergence of its multidrug-resistant isolates in China. Mutations of six of the most frequently reported resistant genes (rpoB, katG, inhA, embB, gyrA, and rpsL) were detected for rifampicin (RIF), isoniazid (INH), ethambutol (EMB), ofloxacin (OFX), and streptomycin (STR) in this study. The amino acid missense mutations (MMs) and their corresponding single nucleotide polymorphism mutations for all drug-resistant (DR) isolates are described in detail. All isolates were divided into non-extensively drug-resistant (Non-XDR) and preXDR/XDR groups. No statistical differences were detected among MMs and linked MMs (LMs) between the two groups, except for rpsL 88 (p = 0.037). In the preXDR/XDR group, the occurrence of MMs in rpoB, katG, and inhA developed phenotypic resistance and MMs of rpoB 531, katG 315, rpsL 43, and rpsL 88 could develop high levels of DR. It is necessary to carry out epidemiological investigations of DR gene mutations in the local region, and thus provide necessary data to support the design of new technologies for rapid detection of resistant M. tb and the optimization of detection targets.

Sulfuric waters and mud in the Tatrespublika

At the moment, when the question of the accelerated development of local resorts is on the turn of the socialist construction of health care, it seems to us quite appropriate to highlight some data concerning the mineral resources of the local region, which could be the basis of resort construction in the Tatrespublika.