A5101960933- Video Surveillance and Tracking Methods
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Wireless Networks and Protocols
- Advanced Image Fusion Techniques
- Remote-Sensing Image Classification
- Mobile Agent-Based Network Management
- Infrared Target Detection Methodologies
- Distributed and Parallel Computing Systems
- Advanced Image and Video Retrieval Techniques
- IPv6, Mobility, Handover, Networks, Security
- Network Traffic and Congestion Control
- Wireless Communication Networks Research
- Human Pose and Action Recognition
- Simulation and Modeling Applications
- Energy Efficient Wireless Sensor Networks
- Mobile Ad Hoc Networks
- Insect and Arachnid Ecology and Behavior
- Visual Attention and Saliency Detection
- Cognitive Radio Networks and Spectrum Sensing
- Vehicle License Plate Recognition
- Industrial Technology and Control Systems
- Insect-Plant Interactions and Control
- Advanced Sensor and Control Systems
- Wireless Body Area Networks
South China University of Technology
2017-2024
Guilin University of Technology
2023-2024
Xi'an Shiyou University
2023
Inner Mongolia University
2012-2023
Shanghai Artificial Intelligence Laboratory
2021-2022
Group Sense (China)
2019-2022
ShangHai JiAi Genetics & IVF Institute
2022
Beihang University
2001-2021
Nanchang Hangkong University
2010-2021
West China Medical Center of Sichuan University
2020-2021
Existing semantic segmentation works mainly focus on learning the contextual information in high-level features with CNNs. In order to maintain a precise boundary, low-level texture are directly skip-connected into deeper layers. Nevertheless, not only about local structure, but also include global statistical knowledge of input image. this paper, we fully take advantages and propose novel Statistical Texture Learning Network (STL-Net) for segmentation. For first time, STL-Net analyzes...
Unsupervised learning has been popular in various computer vision tasks, including visual object tracking. However, prior unsupervised tracking approaches rely heavily on spatial supervision from templatesearch pairs and are still unable to track objects with strong variation over a long time span. As unlimited self-supervision signals can be obtained by video along cycle time, we investigate evolving Siamese tracker videos forward-backward. We present novel framework, which learn temporal...
Background: Bone age can reflect the true growth and development status of a child; thus, it plays critical role in evaluating endocrine disorders. This study established validated an optimized Tanner-Whitehouse 3 artificial intelligence (TW3-AI) bone assessment (BAA) system based on convolutional neural network (CNN).
In order to investigate the genetic diversity of rhizobia associated with various exotic and invasive species in tropical mainland China, 116 bacterial isolates were obtained from Mimosa root nodules collected Sishuangbanna Yuanjiang districts Yunnan province. Isolated characterized by RFLP analysis 16S rRNA genes, SDS-PAGE whole-cell proteins BOX-PCR. Most isolated strains identified as β-rhizobia belonging diverse populations Burkholderia Cupriavidus, phylogenetic relationships their gene...
Lepidopterans play an important role in human economy, since some of them are harmful to vegetation agriculture and others produce useful materials such as silks, etc. To recognize lepidopteran species correctly is very meaningful farmers, forest workers, or even insect researchers. This study proposed a cascade architecture which combines the methods deep convolutional neural network (DCNNs) Supported Vector Machines (SVMs) identify Lepidoptera from their images. The data-set used this...
The Platonic Representation Hypothesis suggests a universal, modality-independent reality representation behind different data modalities. Inspired by this, we view each neuron as system and detect its multi-segment activity under various peripheral conditions. We assume there's time-invariant for the same neuron, reflecting intrinsic properties like molecular profiles, location, morphology. goal of obtaining these neuronal representations has two criteria: (I) segments from should have more...
In order to meet the needs of process inspection technology for industrial equipment, image recognition based on deep learning has shown great potential in field welding defects. this paper, an improved YOLOv8 algorithm is proposed improve defect identification ability workpiece. Through experimental verification selected data sets kaggle, study evaluates detection performance that integrates SCConv C2f module at Backbone level. The results show accuracy compared with traditional version,...
Divisible applications are a class of tasks whose loads can be partitioned into some smaller fractions, and each part executed independently by processor. A wide variety divisible have been found in the area parallel distributed processing. This paper addresses problem how to partition allocate available resources mobile edge computing environments with aim minimizing completion time applications. theoretical model was proposed for partitioning an entire application according load...
Vision sensor systems (VSS) are widely deployed in surveillance, traffic and industrial contexts. A large number of images can be obtained via VSS. Because the limitations vision sensors, it is difficult to obtain an all-focused image. This causes difficulties analyzing understanding In this paper, a novel multi-focus image fusion method (SRGF) proposed. The proposed uses sparse coding classify focused regions defocused focus feature maps. Then, guided filter (GF) used calculate score An...
Among tasks related to intelligent interpretation of remote sensing data, scene classification mainly focuses on the holistic information entire scene. Compared pixel-level or object-based tasks, it involves a richer semantic context, making more challenging. With rapid advancement deep learning, convolutional neural networks (CNNs) have found widespread applications across various domains, and some work has introduced them into tasks. However, traditional convolution operations involve...
There are many decisions which usually made heuristically both in single object tracking (SOT) and multiple (MOT). Existing methods focus on tackling decision-making problems special tasks without a unified framework. In this paper, we propose decision controller (DC) is generally applicable to SOT MOT tasks. The learns an optimal policy with deep reinforcement learning algorithm that maximizes long term performance supervision. To prove the generalization ability of DC, apply it challenging...
Infrared and visible images play an important role in transportation systems since they can monitor traffic conditions around the clock. However, are susceptible to imaging environments, infrared not rich enough detail. The fusion techniques fuse these two different modal into a single image with more useful information. In this paper, we propose effective method for systems. weight maps measured by utilizing sparse coefficients. next is decompose pair high-frequency layers (HFLs)...
Abstract Image fusion technology combines information from different source images of the same target and performs extremely effective complementation, which is widely used for transportation field, medicine surveillance field. Specifically, due to limitation depth field in imaging device, cannot focus on all objects miss partial details. To deal with this problem, an multi-focus image method proposed paper. We interpret production map as a two-class classification task solve problem by...
We describe a new species of forest bark beetle, Tomicus armandii Li & Zhang, collected from Pinus in Yunnan, China. used the D2 fragment 28S rDNA to improve taxonomy Tomicus. The can easily be distinguished other using following two morphological characters: punctures interstria 2 on declivity appearing evenly biseriate or triseriate; erect interstrial setae short, about 0.5× as long distance between striae. genetic distances measured T. and are similar species, these much higher than...