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.

Improving assessment of urban racial segregation by partitioning a region into racial enclaves

Frequently, a single-value metric is needed to rank urban regions with respect to the level of multiracial segregation or to compare a segregation level of a single urban region at two different times. Assessment of segregation depends not only on a metric used but also on a choice of region’s partitioning. The standard practice is to partition the region into single-scale subregions. In the United States, census tracts are the subregions of choice. Census aggregation units including tracts are delineated without direct regard to racial homogeneity and are in fact heterogeneous. Consequently, using tracts as subdivisions leads to the underestimation of the segregation level of the entire region. Here we propose to partition a region into racial enclaves - units having boundaries that align with transitions between different racial compositions. By reflecting true demographic structure, such units minimize their internal racial inhomogeneity resulting in improved assessment of segregation. Enclaves are defined as aggregates of adjacent census blocks (smallest and the most racially homogeneous census units) of similar composition. In a typical US urban region effective population size of enclaves is an order of magnitude larger than the size of a census tract and yet the segregation calculated based on enclaves is larger than segregation based on census tracts. The proposed methodology is described and applied to a set of 61 largest cities in the U.S. in their metropolitan statistical areas (MSAs) as well as their urban areas (UAs) boundaries using 1990 and 2010 block-level data. The method is compared to the standard methodology using correlations between cities’ segregation rankings.

Fusion of multi-scale bag of deep visual words features of chest X-ray images to detect COVID-19 infection

Chest X-ray (CXR) images have been one of the important diagnosis tools used in the COVID-19 disease diagnosis. Deep learning (DL)-based methods have been used heavily to analyze these images. Compared to other DL-based methods, the bag of deep visual words-based method (BoDVW) proposed recently is shown to be a prominent representation of CXR images for their better discriminability. However, single-scale BoDVW features are insufficient to capture the detailed semantic information of the infected regions in the lungs as the resolution of such images varies in real application. In this paper, we propose a new multi-scale bag of deep visual words (MBoDVW) features, which exploits three different scales of the 4th pooling layer’s output feature map achieved from VGG-16 model. For MBoDVW-based features, we perform the Convolution with Max pooling operation over the 4th pooling layer using three different kernels: $$1 \times 1$$ 1 × 1 , $$2 \times 2$$ 2 × 2 , and $$3 \times 3$$ 3 × 3 . We evaluate our proposed features with the Support Vector Machine (SVM) classification algorithm on four CXR public datasets (CD1, CD2, CD3, and CD4) with over 5000 CXR images. Experimental results show that our method produces stable and prominent classification accuracy (84.37%, 88.88%, 90.29%, and 83.65% on CD1, CD2, CD3, and CD4, respectively).

The Development of Multi-scale Data Management for CityGML-based 3D Buildings

The CityGML model is now the norm for smart city or digital twin city development for better planning, management, risk-related modelling and other applications. CityGML comes with five levels of detail (LoD), mainly constructed from point cloud measurements and images of several systems, resulting in a variety of accuracies and detailed models. The LoDs, also known as pre-defined multi-scale models, require large storage-memory-graphic consumption compared to single scale models. Furthermore, these multi-scales have redundancy in geometries, attributes, are costly in terms of time and workload in updating tasks, and are difficult to view in a single viewer. It is essential for data owners to engage with a suitable multi-scale spatial management solution in minimizes the drawbacks of the current implementation. The proper construction, control and management of multi-scale models are needed to encourage and expedite data sharing among data owners, agencies, stakeholders and public users for efficient information retrieval and analyses. This paper discusses the construction of the CityGML model with different LoDs using several datasets. A scale unique ID is introduced to connect all respective LoDs for cross-LoD information queries within a single viewer. The paper also highlights the benefits of intermediate outputs and limitations of the proposed solution, as well as suggestions for the future.

IGBT Open-Circuit Fault Diagnosis for MMC Submodules Based on Weighted-Amplitude Permutation Entropy and DS Evidence Fusion Theory

Modular multilevel converters (MMCs) have a complex structure and a large number of submodules (SMs). If there is a fault in one of the SMs, it will affect the reliable operation of the system. Therefore, rapid fault diagnosis and accurate fault positioning are crucial to ensuring the continuous operation of the system. However, the IGBT open-circuit faults in different submodules of MMCs have similar fault features, and the traditional fault feature extraction method cannot effectively extract the key features of the fault so as to accurately locate the faulty submodules. In response to this problem, this paper proposes a fault diagnosis method based on weighted-amplitude permutation entropy (WAPE) and DS evidence fusion theory. The simulation results show that WAPE has better feature extraction ability than basic permutation entropy, and the fused multiscale feature decision output has better diagnostic accuracy than the single-scale feature. Compared with traditional fault diagnosis methods, this method achieves the diagnosis of multiple fault types by collecting a single signal, which greatly reduces the number of samples and leads to higher diagnostic accuracy and faster diagnostic speed.

Issues in specifying contrast in building elements for people with a visual disability

BACKGROUND: Standards writers, national and international, have used different contrast calculations to set requirements in building elements for people with visual impairments. On the other hand, they have typically set a single requirement (30%) for specifying the minimum contrast. The systems are not linearly related and 30%means something rather different in each system. OBJECTIVE: To provide a comparison of the various scales in order to illustrate the differences caused by multiple scales with a single compliance value, recommend a single scale for universal adoption and, if a new measure is problematic for implementation, to recommend the most perceptually uniform of the present methods. METHODS: We use the contrast between combinations of 205 paint colours to illustrate the relationships between the measures. We use an internationally accepted scale, with equal perceptual steps, as a “gold standard” to identify the most perceptually uniform measurement scale in the existing methods. RESULTS: We show that Michelson contrast is the most perceptually uniform of the existing measurement scales. We show the contrasts in the proposed method that equate to the various current requirements. CONCLUSIONS: We propose that CIE Metric Lightness could be used as the contrast measure. Alternatively, Michelson contrast is the most perceptually linear of the current measurement scales.

A New Anelasticity Model for Wave Propagation in Partially Saturated Rocks

Elastic wave propagation in partially saturated reservoir rocks induces fluid flow in multi-scale pore spaces, leading to wave anelasticity (velocity dispersion and attenuation). The propagation characteristics cannot be described by a single-scale flow-induced dissipation mechanism. To overcome this problem, we combine the White patchy-saturation theory and the squirt flow model to obtain a new anelasticity theory for wave propagation. We consider a tight sandstone Qingyang area, Ordos Basin, and perform ultrasonic measurements at partial saturation and different confining pressures, where the rock properties are obtained at full-gas saturation. The comparison between the experimental data and the theoretical results yields a fairly good agreement, indicating the efficacy of the new theory.

The kapok petal: superhydrophobic surface induced by microscale trichomes

The Kapok petal is reported for the first time that it shows a superhydrophobic characteristic with a static water contact angle higher than 150°. Intriguingly, there exist single-scale micro-trichomes and no more nanocrystals on a kapok petal in contrast to most natural superhydrophobic surfaces with hierarchical morphologies, such as lotus leaf and rose petal. Experiment results show that kapok petal has an excellent self-cleaning ability either in air or oil. Further scanning electron microscope characterization demonstrates that the superhydrophobic state is induced by densely-distributed microscale trichomes with an average diameter of 10.2 μm and a high aspect ratio of 17.5. A mechanical model is built to illustrate that the trichomes re-entrant curvature should be a key factor to induce the superhydrophobic state of the kapok petal. To support the proposed mechanism, gold-wire trichomes with a re-entrant curvature are fabricated and the results show that a superhydrophobic state can be induced by microstructures with a re-entrant curvature surface. Taking the scalability and cost-efficiency of microstructure fabrication into account, we believe the biomimetic structures inspired by the superhydrophobic kapok petal can find numerous applications that require a superhydrophobic state.

A Modular Workflow for Model Building, Analysis, and Parameter Estimation in Systems Biology and Neuroscience

Neuroscience incorporates knowledge from a range of scales, from single molecules to brain wide neural networks. Modeling is a valuable tool in understanding processes at a single scale or the interactions between two adjacent scales and researchers use a variety of different software tools in the model building and analysis process. Here we focus on the scale of biochemical pathways, which is one of the main objects of study in systems biology. While systems biology is among the more standardized fields, conversion between different model formats and interoperability between various tools is still somewhat problematic. To offer our take on tackling these shortcomings and by keeping in mind the FAIR (findability, accessibility, interoperability, reusability) data principles, we have developed a workflow for building and analyzing biochemical pathway models, using pre-existing tools that could be utilized for the storage and refinement of models in all phases of development. We have chosen the SBtab format which allows the storage of biochemical models and associated data in a single file and provides a human readable set of syntax rules. Next, we implemented custom-made MATLAB® scripts to perform parameter estimation and global sensitivity analysis used in model refinement. Additionally, we have developed a web-based application for biochemical models that allows simulations with either a network free solver or stochastic solvers and incorporating geometry. Finally, we illustrate convertibility and use of a biochemical model in a biophysically detailed single neuron model by running multiscale simulations in NEURON. Using this workflow, we can simulate the same model in three different simulators, with a smooth conversion between the different model formats, enhancing the characterization of different aspects of the model.