A5101562421- Infrastructure Maintenance and Monitoring
- Railway Engineering and Dynamics
- Advancements in Semiconductor Devices and Circuit Design
- Vehicle License Plate Recognition
- Advanced Measurement and Detection Methods
- Radio Frequency Integrated Circuit Design
- Surface Roughness and Optical Measurements
- Semiconductor materials and devices
- Remote Sensing and LiDAR Applications
- Video Surveillance and Tracking Methods
- Optical measurement and interference techniques
- Advanced Neural Network Applications
- Hand Gesture Recognition Systems
- Anomaly Detection Techniques and Applications
- Hydrogen embrittlement and corrosion behaviors in metals
- Advanced Measurement and Metrology Techniques
- Image Enhancement Techniques
- Industrial Vision Systems and Defect Detection
- Non-Destructive Testing Techniques
- Acoustic Wave Phenomena Research
- Silicon Carbide Semiconductor Technologies
- Video Analysis and Summarization
- Structural Health Monitoring Techniques
- Advanced Vision and Imaging
- Advanced Sensor and Control Systems
China Academy of Railway Sciences
2016-2024
Wuhan Polytechnic University
2024
Hunan Normal University
2011-2022
Changsha Normal University
2022
Beijing Jiaotong University
2012-2020
China Special Equipment Inspection and Research Institute
2020
East China University of Science and Technology
2015-2017
North China University of Science and Technology
2015-2017
Shandong Jianzhu University
2015-2017
Shandong University
2007-2016
Vision-based automatic railway fastener inspection, instead of manual operation, remains a great challenge. Even though many supervised learning-based methods have been developed, expensive training labels and imbalanced data are the main obstacles to leverage performance inspection task. To tackle problems, we present novel vision-based system (VFIS) which is inspired by few-shot learning. VFIS can automatically collect annotate large number samples using proposed online template...
At present, the method of two-dimensional image recognition is mainly used to detect abnormal fastener in rail-track inspection system. However, too-tight-or-too-loose condition may cause clip break or loose due high frequency vibration shock, which difficult from image. In this practical application background, 3D visual detection technology provides a feasible solution. paper, we propose fundamental multi-source data method, as well an accurate and robust location nut bolt segmentation...
Along with the rapid development of connected vehicle communication technology, describing following driving status becomes gradually complicated. Driver behavior, type, and road factors affect speed, distance reflects variability. In this paper, a nonlinear model is constructed to characterize The based on full speed difference (FVD), introduces headway time coefficient, type advance response parameter reflecting driver's personal characteristics, slope coefficient curve curvature...
Vision-based abnormal object detection in railway track inspection images is one of the critical tasks to ensure safety transportation. Even though many machine learning-based methods have been developed, these approaches heavily rely on anomaly supervisions and therefore cannot detect unknown classes. In order tackle problem, this article proposes a novel unsupervised method objects, which does not require training data. Specially, we find that image almost symmetrical about central line,...
Current state-of-the-art intensity and depth modal-based railway surface defect inspection system faces the dilemma between false alarming miss detection. To overcome this challenge, a bimodal detection scheme using both feature-level fusion (evidence theory-based) decision-level is designed. Moreover, an improved evidential algorithm proposed, which adopts three-branched weight structure introduces Transferable Belief Model to decision functions, achieving outstanding performance on...
Abstract Rail surface defects are potential danger factors for railway systems, and visual inspection of plays a vital role in rail maintenance. Recently, the methods based on deep learning have been widely used but such systems often face problem lack defect samples training models at start‐up, which is called cold‐start problem. It challenging to obtain sufficient since sparse even non‐existent running system. Therefore, synthetic‐to‐realistic domain adaptation (SRDA) method proposed...
Current single-structured light sensor-based rail-track section identification lacks robustness against unstable signal transmission. The decision-level sensor fusion based on evidence theory is fragile to conflict. Aiming at the listed challenges, this study proposes a robust scheme that combines multistructured sensors and new evidence-theoretic strength of kernel method. In scheme, multiple are involved tackle lack robustness, kernel-induced belief metric (KIBM), which first connects...
Abstract As a critical component of ensuring the safe and stable operation trains, detection bird’s nests on rail catenary has always been essential. Low-resolution images lack labelled data, however, make it difficult to detect smaller (those occupying small pixels in input image). Previous solution relies manual online patrol or offline video playback, which severely limits efficiency. Previously, this challenge was addressed by We propose work context-guided coarse-to-fine model (CG-CFDM)...
We have investigated the radio frequency (RF) extrinsic resistance extraction for partially-depleted (PD) silicon-on-insulator (SOI) metal-oxide-semiconductor field effect transistors (MOSFETs). Although thick buried oxide in SOI devices can block substrate coupling, neutral-body coupling is significant RF applications. An equivalent circuit considering this has been proposed. Based on circuit, a new model capturing dependence of resistances derived. After impact quasi-neutral body, we...
The reflection characteristics of the unit cell, consisting a subwavelength circular hole and rigid wall, was discussed theoretically, it found that phase shift reflected waves could cover almost 2π span by adjusting radius when acoustic normally impinge on it. Based analytical formulas, an metasurface (AMS) sample constructed array cells with different radii designed fabricated. sound pressure fields induced were then measured through experimental setup field pattern derived after data...
The problem of foreign object intrusion onto the track bed often occurs in actual operation process high-speed railways. To solve problem, we propose an anomaly detection method for ballastless bed, which is based on semantic segmentation. Firstly, put forward RFODLab segmentation network according to randomness objects distribution, and a small proportion target pixels image. results image obtained through this model can be used obtain accurate pixel information objects. further improve...
The Integrated patrolling inspection train has been used worldwide for railway safety monitoring. camera mounted under the can capture track image abnormal fastener detection. For solving high false positive alarm of rail recognition arising from ballasts occlusion and non-uniform illumination, we proposed a defect method using deep learning model, constructed four network structures based on AlexNet ResNet to learn feature in complex background. experimental results show that RestNet18...
This paper detects violations of smoking in non-smoking areas by construction workers, uses the YOLO object detection algorithm combined with Kalman filter to track human body, then Alphapose's pose estimation obtain key points body. We input into Spatial-Temporal Graph Convolutional Networks for preliminary identification workers' behavior. However, this will lose texture feature information image, resulting drinking, scratching, etc. also be recognized as smoking. Therefore, based on...
The non-coplanar lasers on both sides of the rail during full-section profile measurement based line-structured light vision will cause measured to be distorted, resulting in errors. Currently, field measurement, there are no effective methods for evaluating laser plane attitude, and it is impossible determine degree coplanarity quantitatively accurately. This study proposes an evaluation method fitting planes response this problem. Real-time with three planar targets different heights...