A5100778781- Data Quality and Management
- Advanced Graph Neural Networks
- Video Surveillance and Tracking Methods
- Algorithms and Data Compression
- Anomaly Detection Techniques and Applications
- Hand Gesture Recognition Systems
- History, Medicine, and Leadership
- Data-Driven Disease Surveillance
- Advanced Neural Network Applications
- Computer Graphics and Visualization Techniques
- Face recognition and analysis
- Greenhouse Technology and Climate Control
- Expert finding and Q&A systems
- Ethics in medical practice
- Geochemistry and Geologic Mapping
- Simulation and Modeling Applications
- Chromosomal and Genetic Variations
- Advanced Image and Video Retrieval Techniques
- Multimodal Machine Learning Applications
- Plant Water Relations and Carbon Dynamics
- Quality Function Deployment in Product Design
- Multi-Criteria Decision Making
- Advanced Text Analysis Techniques
- Text and Document Classification Technologies
- Artificial Intelligence in Healthcare
North China University of Technology
2011-2025
Northwest A&F University
2022
China Geological Survey
2014
Background: The main challenge in human–object interaction detection (HOI) is how to accurately reason about ambiguous, complex, and difficult recognize interactions. model structure of the existing methods relatively single, image input may be occluded cannot recognized. Methods: In this paper, we design a Pose-Aware Interaction Network (PAIN) based on transformer architecture human posture address these issues through two innovations: A new feature fusion method proposed, which fuses pose...
Objects thrown from tall buildings in communities are characterized by their small size, inconspicuous features, and high speed. Existing algorithms for detecting such objects face challenges, including excessive parameters, overly complex models that difficult to implement, insufficient detection accuracy. This study proposes a lightweight model communities, named S-YOLOv5, address these issues. The is based on the YOLOv5 algorithm, convolutional neural network, Enhanced ShuffleNet (ESNet),...
This research proposes a face detection algorithm named LighterFace, which is aimed at enhancing speed to meet the demands of real-time community applications. Two pre-trained convolutional neural networks are combined, namely Cross Stage Partial Network (CSPNet), and ShuffleNetv2. Connecting optimized network with Global Attention Mechanism (GAMAttention) extends model compensate for accuracy loss caused by optimizing structure. Additionally, learning rate dynamically updated using cosine...
Simple Sequence Repeats (SSRs) are short tandem repeats of nucleotide sequences. It has been shown that SSRs associated with human diseases and medical relevance. Accordingly, a variety computational methods have proposed to mine from genomes. Conventional rely on high-quality complete genome identify SSRs. However, the sequenced often misses several highly repetitive regions. Moreover, many non-model species no entire With recent advances next-generation sequencing (NGS) techniques,...
Accurate prediction of food-safety risk is an effective measure to improve food prevention. The complex and continuous characteristics safety influence various factors, thus, a model based on temporal knowledge graph proposed in this paper. First, dataset constructed by collecting supervision management sampling data used daily life from 2018 2021 the State Administration for Market Regulation. contains five categories: fruits, vegetables, meat, aquatic products, dairy products. Then, novel...
Combining with virtual reality technology, we establish the multi-user campus roaming system using Lyinux engine, realize interactive roaming, panoramic navigation and real-time chatting. This paper focuses on improving rendering speed of scene implementing synchronization between multi-users. The experiment result shows that this achieves good results in interaction, sense reality.
To greatly support the assessment of food safety situation, Question Answering (QA) system could effectively extract correct answer from web-scale information with relevant questions. However, most existing QA systems are constructed on searching methods, which cannot deal questions that needs multi-hop (multi steps) reasoning process. Meanwhile, black-box models like neural networks base systems, resulting in weak interpretability and low confidence level address these issues, this paper...
In response to the country's call promote modernization of governance with wisdom, data published by information platform is used build a traffic safety knowledge graph, reveal logical relationship between cases road violations, punishment results, and basis, which can quickly effectively assist handling violations. This paper uses Selenium crawler technology obtain Beijing violation dataset from comprehensive service management platform, Jaccard similarity text matching method normalize...
In order to improve the relevance and efficiency of food safety supervision, a knowledge database named entity recognition (NER) model that can extract entities such as poison, disease, symptom has been constructed. First, key content ''Food Safety Accident Determination Prevention Control'' was identified, 8265 valid statements were obtained. It then used data cleaning sequence marking obtain dataset, dividing dataset into training, validation test sets in ratio 8:1:1. The Chinese...