This paper presents a multi-sensor ensemble approach to human physical activity (PA) recognition, using random forest. We designed an learning algorithm, which integrates several independent Random Forest classifiers based on different sensor feature sets build more stable, accurate and faster classifier for recognition. To evaluate the PA data collected from PAMAP (Physical Activity Monitoring Aging People), is standard, publicly available database, was utilized train test. The experimental...

Radar-based sensors do not require optimal lighting and atmospheric conditions nonocclusion, making them a promising solution for human behavior analysis in complex environments. Existing radar-based models generally retrieve features from either the time-velocity domain or time-range domain. Such two-dimensional representations cannot fully depict dynamic motion features. In this paper, we propose temporal range-Doppler PointNet-based method to analyze behavior. We transform echoes 3D point...

Fine-scale land use and cover (LULC) data in a mining area are helpful for the smart supervision of activities. However, complex landscape open-pit areas severely restricts classification accuracy. Although deep learning (DL) algorithms have ability to extract informative features, they require large amounts sample data. As result, design more interpretable DL models with lower demand is highly important. In this study, novel multi-level output-based belief network (DBN-ML) model was...

Retinal blood vessels segmentation plays an important role for retinal image analysis. In this paper, we propose robust vessel method based on reinforcement local descriptions. A novel line set feature is firstly developed to capture shape information of by employing the length prior vessels, which intensity variety. After that, calculated each pixel, and then morphological gradient extracted enhancing edge smaller vessel. At last, feature, are combined obtain Compared with existing...

Groups are the basic elements which depict complex social dynamics in crowded scenarios. Most researchers tend to partition people based on trajectories and lack sufficient utilization of group properties. To address this issue, we design a new detection framework multi-level descriptors. Both crowd semantic information utilized profile A unified graph clustering model is given multi-feature consistency inconsistency objective function. The public pedestrian datasets show that proposed can...

In central receiver systems, the ideal reflective shape of a heliostat is section paraboloid that adapting with sun's angle and mirror's location in field. Deviation from this leads to optical astigmatism increases spot size on aperture, which eventually causes higher energy loss lower conversion efficiency. However, it challenging implement by conventional design manufacturing methods. paper, novel compliant methodology proposed. By tailoring two dimensional stiffness profile square plate,...

Over the past few decades, low recovery of resource coupled with strong strata behavior has plagued development mines in Western China. To address these issues, an emerging and innovative technology- roadway backfill mining method was proposed this paper on basis compressive properties materials. A rational composition for materials is obtained layout technology are presented. Once placed, together barrier pillar have to satisfy certain mechanical stability requirement ensure a safe...

Activity recognition based on mobile devices has been used in a wide range of applications, including health monitoring, context-aware and inertial positioning. However, the single-layer classifier model cannot accurately recognize users' physical activity due to diversity activities. This paper described evaluated with hierarchical classifiers for recognition. To implement model, reasonable effective pattern combination algorithm similarity between activities was put forward design...

Abstract Cyclist trajectory prediction is an essential task in autonomous driving and surveillance systems. This challenging due to that the bicycles go much faster than pedestrians a minor error could lead severe deviation actual path. Existing cyclist models usually employ social pooling mechanism depict mutual interactions between targets. They ignore operation leaky information. Moreover, they prefer use recurrent architecture capture time‐varying features, which not efficient...

MIG welding and laser-MIG hybrid have been widely used to joint aluminum alloy in recent years. Residual stress heat cycling of are analyzed by SYSWELD software. The results show that the peak values is 30~50% less than welded joints.

In a Solar Power Tower (SPT) system, the ideal shape of heliostat concentrator is section paraboloid which function location in array and incidence sun angle. This difficult to achieve limits system efficiency. A shape-optimized compliant (SOC) design parabolic heliostats presented here solve this problem. An approximation suggested use an optimized stationary only varies with array. structure proposed that simple flat mirror two-dimensional tailored stiffness profile form required surface...

Active fault is the most dangerous natural hazards of buried steel pipelines, as large stress and strain induced by ground movement can lead to pipe failure, which may cause severe accidents. Based on nonlinear finite element method, local buckling behavior high strength X80 pipelines under compression strike-slip was studied systematically. Accuracy numerical model validated previous full scale experimental results. A baseline analysis performed elucidate phenomenon pipe. Parametric also...

Music learning plays an important role in cultivating the aesthetic taste and developing creativity of young people. The current traditional music education is unable to provide rich diverse resources, nor can it pay attention individual differences among people achieve personalized teaching. This article conducted extensive research on development a intelligent system by combining cloud computing improved ant colony optimization (ACO). goal was improve quality for teens accomplish...

The mechanical properties of cemented paste backfill (CPB) directly determine the effect backfilling in coal mining. However, current studies on CPB are all based standard cylinder or cube specimens, which cannot accurately characterise distribution different spatial positions situ. In addition, high-water-content material (HWM) has developed rapidly as a new binder recent years. Therefore, this paper, uniaxial compressive strength (UCS) and elastic modulus with gangue aggregate HWM...