In recent years, deep learning based methods have achieved promising performance in standard object detection. However, these lack sufficient capabilities to handle underwater detection due challenges: (1) Objects real applications are usually small and their images blurry, (2) the datasets accompany heterogeneous noise. To address two problems, we first propose a novel neural network architecture, namely Sample-WeIghted hyPEr Network (SWIPENet), for SWIPENet consists of high resolution...

Point cloud semantic segmentation (PCSS) is crucial for digital twins of high-speed railways. By now, the concerned subjects are confined within interior infrastructures However, surrounding environments also important safe operation. Concerning this issue, a full-range railway scanning scheme based on unmanned-aerial-vehicle-borne LiDAR utilized. massive data volume and distribution imbalance pose great challenges PCSS. To address these issues, novel PCSS framework called 3DGraphSeg...

The height and stagger of the contact wire directly affect energy supply high-speed trains. To ensure operation safety, there is an urgent demand for railways to measure static parameters wires all over line with high precision efficiency. However, this issue barely discussed. Concerning issue, paper proposes a UAV-LiDAR-based measuring framework railway wire. By mounting LiDAR on UAV, can efficiently collect data from lines in service without occupying train operating-diagrams. It extremely...

Abstract Autonomous railway inspection with unmanned aerial vehicles (UAVs) has huge advantages over traditional methods. As a prerequisite for UAV‐based autonomous following of lines, it is quite essential to develop intelligent track detection algorithms. However, there are no existing algorithms currently that can efficiently adapt the demand various forms and changing inclination angles tracks in UAV images. To address challenge, this paper proposes novel anchor‐adaptive network...

Advancements in artificial intelligence (AI) have greatly benefited plant phenotyping and predictive modeling. However, unrealized opportunities exist leveraging AI advancements model parameter optimization for fitting complex biophysical models. This work developed novel software, PhoTorch, parameters of the Farquhar, von Caemmerer, Berry (FvCB) biochemical photosynthesis based components popular framework PyTorch. The primary novelty software lies its computational efficiency, robustness...

The cross-border port serves as a crucial travel connecting mainland China with Hong Kong and Macau, directly impacting the overall satisfaction of travel. While previous studies on neighborhoods, communities, other areas have thoroughly examined heterogeneity asymmetry in satisfaction, research at ports remains notably limited. This paper explores using gradient boosted decision trees (GBDT) k-means cluster analysis under framework three-factor theory, aiming to demonstrate latest...

The cross-border port serves as a crucial travel connecting mainland China with Hong Kong and Macau, directly impacting the overall satisfaction of travel. While previous studies on neighborhoods, communities, other areas have thoroughly examined heterogeneity asymmetry in satisfaction, research at ports remains notably limited. This paper explores using gradient boosted decision trees (GBDT) k-means cluster analysis under framework three-factor theory, aiming to demonstrate latest...

UAV-based automatic railway inspection is expected to have the potential reform of railways. In this area, real-time scene parsing quite essential. However, limited computation resources UAV onboard computer pose a huge challenge for algorithm juggle precise prediction with strong timeliness. Concerning issue, paper proposes novel named deep fully decoupled residual convolutional network, which consists blocks (Non-bottleneck-FDs) deal dilemma between high demand and resources. The block...

Point cloud semantic segmentation for railway infrastructures is an essential step towards establishing digital twins. Deep learning-based methods have shown great potential in this field compared to traditional that rely on hand-crafted features. However, deep point clouds still face typical challenges need be addressed. In regard, we propose a novel learning framework named SALAProNet, which consists of set effective and concise modular solutions. The first challenge addressed the massive...

Lithium-ion battery manufacturing is a highly complicated process with strongly coupled feature interdependencies, feasible solution that can analyse variables within chain and achieve reliable classification thus urgently needed. This article proposes random forest (RF)-based framework, through using the out of bag (OOB) predictions, Gini changes as well predictive measure association (PMOA), for effectively quantifying importance correlations features their effects on electrode properties....

The dynamic input-output model with coal mine safety was established on the basis of table in this paper. And paper gives definitions and proposes methods to make sure investment coefficient productive, security, consumption accident. can be solved reverse recursive solution, applied analysis input- output for forecast input.

In order to study the application of computer digital image processing technology in film and television (FAT) animation visual sensing expression, by studying principle technology, a spatial adaptive steganography enhancement algorithm multiscale filters is proposed carry out original FAT production. This can provide more high‐quality refined materials for production, which convenient postproduction produce higher‐resolution clear works. Finally, verified. The results show that has high...

UAVs have a broad application prospect in the field of railway inspection due to their excellent mobility and flexibility. However, it still faces challenges, such as high human labor costs low intelligence levels. Therefore, is great significance develop real-time intelligent rail recognition algorithm that can be deployed on onboard computing device guide UAV's camera follow target area complete automatically. significant issue rails from perspective may appear with changing pixel widths...

In recent years, deep learning based methods have achieved promising performance in standard object detection. However, these lack sufficient capabilities to handle underwater detection due challenges: (1) Objects real applications are usually small and their images blurry, (2) the datasets accompany heterogeneous noise. To address two problems, we first propose a novel neural network architecture, namely Sample-WeIghted hyPEr Network (SWIPENet), for SWIPENet consists of high resolution...

Simulating soil reflectance spectra is invaluable for soil-plant radiative modeling and training machine learning models, yet it difficult as the intricate relationships between structure its constituents. To address this, a fully data-driven optics generative model (SOGM) simulation of based on property inputs was developed. The trained an extensive dataset comprising nearly 180,000 spectra-property pairs from 17 datasets. It generates text-based describing properties their values rather...

Deep neural network (DNN) is extensively explored for LiDAR-based 3D object detection, a crucial perception task in the field of autonomous driving. However, presence redundant parameters and complex computations pose challenges practical deployment DNNs. Despite knowledge distillation (KD) an effective approach accelerating models, extremely small number efforts explore its potential on LiDARbased detectors. Besides, existing studies neglect to elaborately investigate voxel-wise features...

The rail fastener is an important infrastructure of railway line system to ensure the safety operation. There urgent need for a set automatic defect inspection scheme fasteners which have high efficiency and accuracy. To this end, paper proposes modified lightweight YOLOv5 model considering application scenario UAV. We reconstruct backbone on basis ShuffleNetV2 RepVGG, switch detector head decoupled type from YOLOX. Data augmentation adopted address problem deficient samples. results show...

With the rapid development of oil and gas pipeline construction, there are more pipelines have to pass through mineral deposits area. The surface cracking, subsidence collapse will be occurred commonly because underground goaf after exploitation in area, pipelines’ safety is faced with hidden troubles. In order solve contradiction between construction minerals mining, some research on design method mined-out area was carried out. prediction constituted influence factors analysis, deformation...

The Beijing-Shanghai High speed railway line (Hereinafter referred to as “Jing-Hu HSL”) is one of the most important lines in Chinese rapid passenger transportation network and will be put into operation at end 2011. Train planning directly reflects quality competition ability train services. characteristics operational conditions flow this corridor HSL bring about a few new issues on like night operation, OD sets, cyclic stop schedule. For first issue, large amount long distance travel...