Weidong Min

ORCID: 0000-0003-2526-2181
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Video Surveillance and Tracking Methods
  • Advanced Neural Network Applications
  • Human Pose and Action Recognition
  • Advanced Image and Video Retrieval Techniques
  • Anomaly Detection Techniques and Applications
  • Gait Recognition and Analysis
  • Face and Expression Recognition
  • Face recognition and analysis
  • Context-Aware Activity Recognition Systems
  • Vehicle License Plate Recognition
  • Hand Gesture Recognition Systems
  • Digital Media Forensic Detection
  • Generative Adversarial Networks and Image Synthesis
  • Advanced Steganography and Watermarking Techniques
  • Remote-Sensing Image Classification
  • Image Retrieval and Classification Techniques
  • Advanced Image Processing Techniques
  • Fire Detection and Safety Systems
  • Biometric Identification and Security
  • Image Processing Techniques and Applications
  • Image Enhancement Techniques
  • AI in cancer detection
  • 3D Shape Modeling and Analysis
  • Machine Learning and ELM
  • Domain Adaptation and Few-Shot Learning

Nanchang University
2016-2025

Jiangxi University of Technology
2023-2024

Education Department of Jiangxi Province
2019-2022

Tianjin University
2015

Tsinghua University
1993-2015

Tiangong University
2011-2015

University of Alberta
2015

Nankai University
2015

Sign language recognition aims to recognize meaningful movements of hand gestures and is a significant solution in intelligent communication between the deaf community hearing societies. However, until now, current dynamic sign methods still have some drawbacks with difficulties recognizing complex gestures, low accuracy for most recognition, potential problems larger video sequence data training. In order solve these issues, this paper presents multimodal method based on deep 3-dimensional...

10.1109/access.2019.2904749 article EN cc-by-nc-nd IEEE Access 2019-01-01

Successful remote sensing image registration is an important step for many applications. The scale-invariant feature transform (SIFT) a well-known method registration, with variants of SIFT proposed. However, it only uses local low-level information, and loses much middle- or high-level information to register. Image features extracted by convolutional neural network (CNN) have achieved the state-of-the-art performance classification retrieval problems, can provide registration. Hence, in...

10.1109/lgrs.2017.2781741 article EN IEEE Geoscience and Remote Sensing Letters 2018-01-04

Various investigations have shown that driver fatigue is the main cause of traffic accidents. Research on use computer vision techniques to detect signs from facial actions, such as yawning, has demonstrated good potential. However, accurate and robust detection yawning difficult because complicated actions expressions drivers in real driving environment. Several same mouth deformation yawning. Thus, a novel approach detecting based subtle action recognition proposed this study alleviate...

10.1109/tmm.2020.2985536 article EN IEEE Transactions on Multimedia 2020-04-06

Remote sensing image retrieval (RSIR) is a fundamental task in remote sensing. Most content-based RSIR approaches take simple distance as similarity criteria. A method based on weighted and basic features of convolutional neural network (CNN) proposed this letter. The contains two stages. First, offline stage, the pretrained CNN fine-tuned by some labeled images from target data set, then used to extract features, set. Second, online we use model feature query calculate weight each class...

10.1109/lgrs.2018.2847303 article EN IEEE Geoscience and Remote Sensing Letters 2018-07-06

Automatic human fall detection is one important research topic in caring for vulnerable people, such as elders at home and patients medical places. Over the past decade, numerous methods aiming solving problem were proposed. However, existing only focus on detecting themselves cannot work effectively complicated environments, especially falls furniture. To alleviate this problem, a new method furniture using scene analysis based deep learning activity characteristics presented paper. The...

10.1109/access.2018.2795239 article EN cc-by-nc-nd IEEE Access 2018-01-01

Traffic sign recognition (TSR) plays an important role in driving assistance system and traffic safety insurance. However, existing methods focus on extracting features of signs ignore the constraints spatial positional relationships between other objects scene. This way results incorrectly detecting similar as failing to detect very small signs. A TSR method based semantic scene understanding structural location is proposed this study solve aforementioned problems. structure model establish...

10.1109/tits.2022.3145467 article EN IEEE Transactions on Intelligent Transportation Systems 2022-02-01

Semantic segmentation for large-scale point clouds in 3D computer vision remains challenging. Most existing studies focus on creating complex local geometry extractors without considering the sparsity of and multi-scale problem objects large-scale, resulting networks that fail to efficiently extract features affect accuracy. In this study, we propose a novel Feature Affine Residual (FA-Res) learnable module learn robust cloud semantic information from domain. First, create Local...

10.1016/j.jag.2023.103259 article EN cc-by-nc-nd International Journal of Applied Earth Observation and Geoinformation 2023-03-20

It plays an important role to accurately track multiple vehicles in intelligent transportation, especially vehicles. Due complicated traffic environments it is difficult and robustly, when there are occlusions among To alleviate these problems, a new approach proposed with the combination of robust detection two classifiers. An improved ViBe algorithm for accurate uses gray-scale spatial information build dictionary pixel life length make ghost shadows object's residual quickly blended into...

10.1109/tits.2017.2756989 article EN IEEE Transactions on Intelligent Transportation Systems 2017-12-25

Extracting vehicle information is of great significance to the construction Internet Vehicles (IoV). Vehicle logo detection (VLD) technology can effectively extract information. Due complex traffic environment, existing methods have difficulty accurately detect logo, especially when has motion blur. To alleviate these problems, a new approach proposed under blur with combination Filter-DeblurGAN and VL-YOLO. possesses judgment mechanism, which determine whether image needs be deblurred based...

10.1109/tvt.2020.2969427 article EN IEEE Transactions on Vehicular Technology 2020-02-02

A comprehensive analysis on the energy-efficient strategy in static Wireless Sensor Networks (WSNs) that are not equipped with any energy harvesting modules is conducted this article. First, a novel generic mathematical definition of Energy Efficiency (EE) proposed, which takes acquisition rate valid data, total consumption, and network lifetime WSNs into consideration simultaneously. To best our knowledge, first time EE mathematically defined. The consumption characteristics each individual...

10.1145/3414315 article EN ACM Transactions on Sensor Networks 2020-10-12

The existing deep learning methods for human fall detection have difficulties to distinguish falls from similar daily activities such as lying down because of not using the 3D network. Meanwhile, they are suitable mobile devices heavyweight and consume a large number memories. In order alleviate these problems, two-stream approach with MobileVGG is proposed in this paper. One stream based on motion characteristics body falls, while other an improved lightweight VGG network, named MobileVGG,...

10.1109/access.2019.2962778 article EN cc-by IEEE Access 2020-01-01

In recent years, no-reference/blind image quality assessment (NR-IQA), as a fundamental but challenging research problem, has been attracting significant attention in the field of digital processing. NR-IQA aims to build computational model quantitatively predict subjective from distorted itself without any reference image. Although great efforts have employed develop various algorithms, due its intrinsic difficulty, issues are still and remain largely unexplored date. this paper, we...

10.1080/02564602.2016.1151385 article EN IETE Technical Review 2016-04-08

Efficiently retrieving synthetic aperture radar (SAR) image is an important yet challenging task in the remote sensing field. Due to shortage of labeled SAR images for fine-tuning convolutional neural network (CNN) models, this letter presents unsupervised domain adaptation model based on CNN learn domain-invariant feature between and optical aerial retrieving, which can alleviate burden manual labeling. We extend a deep novel adversarial by adding discriminator pseudolabel predictor....

10.1109/lgrs.2019.2896948 article EN IEEE Geoscience and Remote Sensing Letters 2019-03-05

Vehicle re-identification (Re-ID) methods often fail to achieve robust performance due insufficient training data and domain diversities. Although state-of-the-art apply image-to-image translation or web augmentation, the construct of new datasets will not only introduce noise, but also undergo a mismatch issue with source domain. Moreover, label noise cross-domain in existing distribution technologies cannot be alleviated. In this paper, multi-domain joint learning inter-domain adaptation...

10.1109/tmm.2021.3104141 article EN IEEE Transactions on Multimedia 2021-08-13

Due to the complexity of medical imaging techniques and high heterogeneity glioma surfaces, image segmentation human gliomas is one most challenging tasks in analysis. Current methods based on convolutional neural networks concentrate feature extraction while ignoring correlation between local global. In this paper, we propose a residual mix transformer fusion net, namely RMTF-Net, for brain tumor segmentation. encoder, encoder including network (RCNN) proposed. The gives an overlapping...

10.3390/brainsci12091145 article EN cc-by Brain Sciences 2022-08-27

Multi-frame human pose estimation has long been an appealing and fundamental issue in visual perception. Owing to the frequent rapid motion occlusion videos, this task is extremely challenging. Current state-of-the-art methods seek model spatiotemporal features by equally fusing each frame local sequence, which weakens target information. In addition, existing approaches usually emphasize more on deep while ignoring detailed information implied shallow feature maps, resulting dropping of...

10.1109/tcsvt.2023.3269666 article EN IEEE Transactions on Circuits and Systems for Video Technology 2023-04-24

Falls sustained by subjects can have severe consequences, especially for elderly persons living alone. A fall detection method indoor environments based on the Kinect sensor and analysis of three‐dimensional skeleton joints information is proposed. Compared with state‐of‐the‐art methods, authors’ provides two major improvements. First, possible activity quantified represented a one‐dimensional float array only 32 items, followed recognition using support vector machine (SVM). Unlike typical...

10.1049/iet-cvi.2018.5324 article EN IET Computer Vision 2018-07-13

In order to build a local electricity market (LEM), community members can trade peer-to-peer (P2P) with their neighbors. This paper proposes Hierarchical Bidding and Transaction Structure based on blockchain (HBTS). First, combined the multi-agents, each microgrid corrects estimated cost probability distribution of other microgrids by Bayesian theorem, making its closer accurate probability. Second, for maximize benefits microgrid, this uses Nash equilibrium in Cournot model find optimal...

10.3390/en12101952 article EN cc-by Energies 2019-05-22
Coming Soon ...