Aiming at the problems of large parameters and low detection speed existing remote sensing road network model, a lightweight (LRSR-net) model is proposed in this paper. In feature encoding stage LRSR-net, an efficient target extraction module dilated joint convolution proposed. This can eliminate loss caused by pooling layer, so that use small amount to learn characteristics targets. addition, decoding stage, damage data structure upsampling operation, The effectively map targets suppress...

Visual question answering by using information from multiple modalities has attracted more and attention in recent years. However, it is a very challenging task, as the visual content natural language have quite different statistical properties. In this work, we present method called Adversarial Multimodal Network (AMN) to better understand video stories for answering. AMN, propose learn multimodal feature representations finding coherent subspace clips corresponding texts (e.g., subtitles...

Abstract Ever since its introduction over five decades ago, geometric solid modeling has been crucial for engineering design purposes and is used in software packages such as computer-aided (cad), manufacturing, engineering, etc. Solid models produced by cad have to transfer information from designers manufacturers. Since the emergence of additive manufacturing (AM), a CAD file can also be directly uploaded three-dimensional (3D) printer production. AM techniques allow complex objects...

Abstract Understanding the process of changes in substrate physical, chemical, and biological properties is critical for vegetation ecosystem reconstruction open‐pit coal mining dumps. However, relationships soil biological, bacterial, fungal characteristics dumps grassland area are still unclear. The purpose this study to understand physic‐chemical, under reclamation time series Soil samples were collected from waste dump reclaimed after 2, 5, 10, 15 years Shengli West No. 2 open pit mine...

Appropriate grazing pressure and fertilizer application of nitrogen (N) phosphorus (P) are effective measures to increase grassland productivity. In this study, we report on the interactive effects intensity soil properties, enzyme characteristics, bacterial community compositions. The experiment was set up in a typical Xilingol, Inner Mongolia, had 12 treatments (CC, CN, CP, CNP, LC, LN, LP, LNP, HC, HN, HP, HNP). These consisted three levels crossed with four treatments: no fertilizer, N P...

Considering the effect of heat input tungsten inert gas (TIG) arc welding for T-joint thin-walled parts aluminum alloy 6061-T6, here, characteristics are analyzed via finite element method. The experiments carried out using scanning electron microscope (SEM), optical (OM), and tensile test specimens to investigate microstructure variation weld zone (WZ), heat-affected (HAZ), base metal (BM), mechanical properties T-welded joint. joint explored assessed combined with in terms yield strength,...

In this paper, a novel real-time structural health monitoring (SHM) system based on streaming data is proposed. contrast to traditional SHM system, the proposed implements series of optimizations for transmission and processing reduce latency better satisfy requirement. The concept watermark in adopted address problem when trigger time window calculation under Moreover, well-designed parallel mechanism used multistage computation requirement stream. A case study which applied Shanghai Tower...

The ∞ux and torque of switched reluctance motor (SRM) have a highly nonlinear functional relationship with rotor position phase current, as consequence the double-salient structure stator pole magnetic saturation, which is di-cult to build an accurate analytic model. In order achieve SRM high-performance control, it necessary model for SRM. On basis adequate precise sample data, by taking advantage neural network that has outstanding mapping capability, this paper adopts Back Propagation...

Based on the GPS trajectory data of a freight enterprise in Dalian, China, this paper studies identification loading and unloading points by clustering algorithm. Firstly, analysing characteristics behaviour, combined with spatial temporal distribution truck data, three characteristic variables number trucks passing through certain place, average speed stay time place are extracted. Then, algorithm visual analysis used to obtain target cluster, POI language geographic information is obtained...

Futures trading analysis plays a pivotal role in the development of macroeconomic policies and corporate strategy planning. High-frequency futures data, typically presented as time series, contain valuable historical patterns. To address challenges such non-stationary modeling prices, we propose novel architecture called FuturesNet, which uses an InceptionTime module to capture short-term fluctuations between ask bid orders, well long-short-term-memory (LSTM) with skip connections long-term...

In modern physical education, data-driven evaluation methods have gradually attracted attention, especially the quantitative prediction of students' sports performance through machine learning model. The purpose this study is to use a variety models regress and predict comprehensive scores in baseball training, so as evaluate effectiveness current teaching put forward targeted training optimization suggestions. We set up model students by collecting many characteristics, such hitting times,...

In modern physical education, data-driven evaluation methods have gradually attracted attention, especially the quantitative prediction of students' sports performance through machine learning model. The purpose this study is to use a variety models regress and predict comprehensive scores in baseball training, so as evaluate effectiveness current teaching put forward targeted training optimization suggestions. We set up model students by collecting many characteristics, such hitting times,...

Visual question answering by using information from multiple modalities has attracted more and attention in recent years. However, it is a very challenging task, as the visual content natural language have quite different statistical properties. In this work, we present method called Adversarial Multimodal Network (AMN) to better understand video stories for answering. AMN, inspired generative adversarial networks, propose learn multimodal feature representations finding coherent subspace...

Abstract Nowadays, bio-inspiration has enhanced the creation of sustainable and innovative solutions to modern engineering problems. Nature could inspire mechanical engineers develop ideas as a great source for multifunctional optimized designs. However, it is very difficult extract desired design knowledge from primarily text-based databases mainly focus on describing biological system. The main objective this study build multi-label classification system classify bio-inspired designs...

Abstract The application of Airborne Time-domain Electromagnetic is developing in recent year, but investigation underground space rarely to see China. An ATEM was conducted Xiongan New area the central part China, which known as one most complex shallow structure underground. plan field work, data processing method and inversion introduced paper. Two years results were compared confirm quality. Ground resistivity discussed Geophysical collected clarifying different kind soil, used for layer...