A5100400130- Visual Attention and Saliency Detection
- Video Surveillance and Tracking Methods
- Image Enhancement Techniques
- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Air Quality and Health Impacts
- Complex Network Analysis Techniques
- Traffic Prediction and Management Techniques
- Air Quality Monitoring and Forecasting
- Anomaly Detection Techniques and Applications
- Advanced Vision and Imaging
- Image and Signal Denoising Methods
- Robotics and Sensor-Based Localization
- Face recognition and analysis
- Data Management and Algorithms
- Sparse and Compressive Sensing Techniques
- Opinion Dynamics and Social Influence
- Atmospheric chemistry and aerosols
- Generative Adversarial Networks and Image Synthesis
- Functional Brain Connectivity Studies
- Medical Image Segmentation Techniques
- Neural dynamics and brain function
- Cell Image Analysis Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Image Processing Techniques
Harbin Engineering University
2024-2025
University of California, Santa Barbara
2025
China Academy of Space Technology
2023-2025
Tianjin University
2022-2025
Chinese Academy of Tropical Agricultural Sciences
2024
Yantai University
2023-2024
Tongji University
2024
Shanghai Jiao Tong University
2016-2024
Tongren Hospital
2024
Washington University in St. Louis
2023-2024
Camouflaged object detection (COD), segmenting objects that are elegantly blended into their surroundings, is a valuable yet challenging task. Existing deep-learning methods often fall the difficulty of accurately identifying camouflaged with complete and fine structure. To this end, in paper, we propose novel boundary-guided network (BGNet) for detection. Our method explores extra object-related edge semantics to guide representation learning COD, which forces model generate features...
Vision transformers have recently shown strong global context modeling capabilities in camouflaged object detection. However, they suffer from two major limitations: less effective locality and insufficient feature aggregation decoders, which are not conducive to camou-flaged detection that explores subtle cues indistinguishable backgrounds. To address these issues, this paper, we propose a novel transformer-based Feature Shrinkage Pyramid Network (FSPNet), aims hierarchically decode...
Spotting camouflaged objects that are visually assimilated into the background is tricky for both object detection algorithms and humans who usually confused or cheated by perfectly intrinsic similarities between foreground surroundings. To tackle this challenge, we aim to extract high-resolution texture details avoid detail degradation causes blurred vision in edges boundaries. We introduce a novel HitNet refine low-resolution representations features an iterative feedback manner,...
The AI City Challenge was created to accelerate intelligent video analysis that helps make cities smarter and safer. Transportation is one of the largest segments can benefit from actionable insights derived data captured by sensors, where computer vision deep learning have shown promise in achieving large-scale practical deployment. 4th annual edition has attracted 315 participating teams across 37 countries, who leverage city-scale real traffic high-quality synthetic compete four challenge...
The rapid advances of transportation infrastructure have led to a dramatic increase in the demand for smart systems capable monitoring traffic and street safety. Fundamental these applications are community-based evaluation platform benchmark object detection multi-object tracking. To this end, we organize AVSS2017 Challenge on Advanced Traffic Monitoring, conjunction with International Workshop Street Surveillance Safety Security (IWT4S), evaluate state-of-the-art tracking algorithms...
When predicting PM2.5 concentrations, it is necessary to consider complex information sources since the concentrations are influenced by various factors within a long period. In this paper, we identify set of critical domain knowledge for forecasting and develop novel graph based model, PM2.5-GNN, being capable capturing long-term dependencies. On real-world dataset, validate effectiveness proposed model examine its abilities both fine-grained influences in process. The PM2.5-GNN has also...
Many real-world ubiquitous applications, such as parking recommendations and air pollution monitoring, benefit significantly from accurate long-term spatio-temporal forecasting (LSTF). LSTF makes use of dependency between spatial temporal domains, contextual information, inherent pattern in the data. Recent studies have revealed potential multi-graph neural networks (MGNNs) to improve prediction performance. However, existing MGNN methods cannot be directly applied due several issues: low...
Depth cues are known to be useful for visual perception. However, direct measurement of depth is often impracticable. Fortunately, though, modern learning-based methods offer promising maps by inference in the wild. In this work, we adapt such models object segmentation using objects' "pop-out" prior 3D. The a simple composition that assumes objects reside on background surface. Such compositional allows us reason about 3D space. More specifically, inferred can localized only information....
The observation-based air pollution forecasting method has high computational efficiency over traditional numerical models, but a poor ability in long-term (after 6 h) due to lack of detailed representation atmospheric processes associated with the transport. To address such limitation, here we propose novel real-time model that applies hybrid graph deep neural network (GNN_LSTM) dynamically capture spatiotemporal correlations among neighborhood monitoring sites better represent physical...
The precuneus/posterior cingulate cortex, which has been associated with pain sensitivity, plays a pivotal role in the default mode network. However, information regarding migraine-related alterations resting-state brain functional connectivity network and local regional spontaneous neuronal activity is not adequate. This study used magnetic resonance imaging to acquire scans 22 migraineurs without aura healthy matched controls. Independent component analysis, data-driven method, was...
With the explosive growth of global spacecraft in orbit, frequency space security events is getting higher and higher, order to guarantee traffic safety a more accurate efficient warning method needed as support. In this paper, algorithm based on telemetry data proposed address problems long cycle time high false alarm rate current routinely applied on-orbit methods. The can effectively control error improve accuracy efficiency by combining orbit maneuvering strategy autonomous planning...
Purpose Visual simultaneous localization and mapping (SLAM) methods suffer from accumulated errors, especially in challenging environments without loop closure. By constructing lightweight offline maps using deep learning (DL)-based technology the two stages, i.e. image retrieval feature matching, goal is to reconstruct six-degree-of-freedom (6-DoF) relationship between SLAM sequences map sequences. This study aims propose a comprehensive coarse-to-fine 6-DoF long-term visual relocalization...
This literature review aims to discuss the application of graph theory in analyzing social and information networks. We first introduce some key network properties, such as clustering coefficient (transitivity), centrality, diameter, which are crucial for understanding dynamics dissemination within Then, we talk about, based on these how can be utilized analyze Lastly, provide an overview various fundamental models including SIR model Linear Threshold model. For model, go over definition...
The proliferation of recommendation systems has revolutionized information retrieval by helping users efficiently navigate through large data sets. However, these often suffer from bias, especially in multimodal setups. This paper addresses the issue bias mitigation using expert systems. Through a comprehensive literature review, various techniques such as knowledge graph integration, fusion, and deep learning architectures are explored. Furthermore, novel approach dynamic meeting algorithms...