Xinyu Yang

ORCID: 0000-0002-3425-0855
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Domain Adaptation and Few-Shot Learning
  • Video Surveillance and Tracking Methods
  • Rock Mechanics and Modeling
  • Human Pose and Action Recognition
  • Geophysical Methods and Applications
  • Music and Audio Processing
  • Digital Media Forensic Detection
  • Food Security and Health in Diverse Populations
  • Landslides and related hazards
  • Advanced Steganography and Watermarking Techniques
  • Music Technology and Sound Studies
  • Homelessness and Social Issues
  • Advanced Neural Network Applications
  • Mental Health Treatment and Access
  • Child Nutrition and Water Access
  • Anomaly Detection Techniques and Applications
  • Education, Safety, and Science Studies
  • Face recognition and analysis
  • Innovative Educational Techniques
  • Advanced Chemical Sensor Technologies
  • Educational Research and Pedagogy
  • Maternal Mental Health During Pregnancy and Postpartum
  • Structural Behavior of Reinforced Concrete
  • LGBTQ Health, Identity, and Policy
  • Advanced Vision and Imaging

Shenzhen University
2014-2024

Lancaster University
2024

University of Bristol
2019-2024

Air Force Engineering University
2024

Taiyuan University of Technology
2021-2023

Cambridge Health Alliance
2021-2023

PAREXEL International (United States)
2023

Beijing Normal University - Hong Kong Baptist University United International College
2022

Civil Aviation University of China
2021

The utilization of deep learning techniques in generating various contents (such as image, text, etc.) has become a trend. Especially music, the topic this paper, attracted widespread attention countless researchers.The whole process producing music can be divided into three stages, corresponding to levels generation: score generation produces scores, performance adds characteristics and audio converts scores with by assigning timbre or generates format directly. Previous surveys have...

10.48550/arxiv.2011.06801 preprint EN other-oa arXiv (Cornell University) 2020-01-01

Abstract We propose a novel end-to-end curriculum learning approach for sparsely labelled animal datasets leveraging large volumes of unlabelled data to improve supervised species detectors. exemplify the method in detail on task finding great apes camera trap footage taken challenging real-world jungle environments. In contrast previous semi-supervised methods, our adjusts parameters dynamically over time and gradually improves detection quality by steering training towards virtuous...

10.1007/s11263-023-01748-3 article EN cc-by International Journal of Computer Vision 2023-01-16

Abstract We present the PanAf20K dataset, largest and most diverse open-access annotated video dataset of great apes in their natural environment. It comprises more than 7 million frames across $$\sim $$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>∼</mml:mo> </mml:math> 20,000 camera trap videos chimpanzees gorillas collected at 18 field sites tropical Africa as part Pan African Programme: The Cultured Chimpanzee. footage is accompanied by a rich set annotations...

10.1007/s11263-024-02003-z article EN cc-by International Journal of Computer Vision 2024-03-04

In order to improve the quality of mathematics education, optimise level textbook writing and promote cultural exchange between China Vietnam, this paper selects junior high school textbooks Zhejiang Education Edition in Kite Vietnam as research objects, uses comprehensive exercise difficulty model compare exercises "triangle" section from both countries. The results show that there is no significant difference overall two versions textbooks, exhibit a distribution characteristic relatively...

10.9734/jamcs/2025/v40i21973 article EN Journal of Advances in Mathematics and Computer Science 2025-02-13

We propose the first multi-frame video object detection framework trained to detect great apes. It is applicable challenging camera trap footage in complex jungle environments and extends a traditional feature pyramid architecture by adding self-attention driven blending both spatial as well temporal domain. demonstrate that this extension can distinctive species appearance motion signatures despite significant partial occlusion. evaluate using 500 videos of apes from Pan African Programme...

10.1109/iccvw.2019.00034 preprint EN 2019-10-01

As an indispensable part of modern human-computer interaction system, speech synthesis technology helps users get the output intelligent machine more easily and intuitively, thus has attracted attention. Due to limitations high complexity low efficiency traditional technology, current research focus is deep learning-based end-to-end which powerful modeling ability a simpler pipeline. It mainly consists three modules: text front-end, acoustic model, vocoder. This paper reviews status these...

10.48550/arxiv.2104.09995 preprint EN cc-by arXiv (Cornell University) 2021-01-01

OBJECTIVES To examine whether Supplemental Nutrition Assistance Program (SNAP) participation is associated with emergency department use among low-income children and any such association mediated by household food hardship child health status and/or moderated special care needs (SHCN) status. We hypothesized SNAP to be reduced likelihoods of use, greater effect sizes for SHCN mediation METHODS In this secondary analysis, we estimated a bivariate probit model (with state-level administrative...

10.1542/peds.2022-058247 article EN PEDIATRICS 2023-01-30

This paper proposes an improved steganalysis algorithm to detect the secret message hidden in compressed video. As majority of video steganographic algorithms modify motion vectors (MV) inter-frame encoding hide data, aliasing effect may be caused distribution difference between MVs two adjacent macroblocks. phenomenon has been observed detecting data that were added cover To exploit correlations neighboring so as more efficiently, we consider joint MV differences one macroblock and other...

10.1109/icip.2014.7026115 article EN 2022 IEEE International Conference on Image Processing (ICIP) 2014-10-01

In this paper we show that learning video feature spaces in which temporal cycles are maximally predictable benefits action classification. particular, propose a novel approach termed Cycle Encoding Prediction (CEP) is able to effectively represent high-level spatio-temporal structure of unlabelled content. CEP builds latent space wherein the concept closed forward-backward as well backward-forward loops approximately preserved. As self-supervision signal, leverages bi-directional coherence...

10.48550/arxiv.2010.07217 preprint EN other-oa arXiv (Cornell University) 2020-01-01

Little is known about depression treatment for transgender and gender diverse (TGD) older adults or TGD people with disabilities. The purpose of this study was to characterize receipt minimally recommended outcomes Medicare beneficiaries.

10.1089/trgh.2022.0146 article EN Transgender Health 2023-03-28

Music generation techniques have made tremendous progress in recent years. Through deep learning neural network frameworks, algorithms can generate music that rivals humans. However, current models lack mainstream and specifications, including algorithm design criteria model evaluation criteria, exactly how to evaluate generated by a is good piece of music. In this case, we did literature review for the three most common currently used generation. These are Biaxial-LSTM, DeepJ MuseGAN. We...

10.1109/icet55676.2022.9824149 article EN 2022 IEEE 5th International Conference on Electronics Technology (ICET) 2022-05-13

We present the PanAf20K dataset, largest and most diverse open-access annotated video dataset of great apes in their natural environment. It comprises more than 7 million frames across ~20,000 camera trap videos chimpanzees gorillas collected at 14 field sites tropical Africa as part Pan African Programme: The Cultured Chimpanzee. footage is accompanied by a rich set annotations benchmarks making it suitable for training testing variety challenging ecologically important computer vision...

10.48550/arxiv.2401.13554 preprint EN cc-by arXiv (Cornell University) 2024-01-01

10.1109/apsipaasc63619.2025.10849068 article EN 2022 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) 2024-12-03

Abstract Failure of residual coal pillars under dynamic load disturbances can induce goaf collapse, ground subsidence, or coalbursts. Encasing the pillar in mortar is an effective method for reinforcing pillar. However, mechanical behaviors mortar-encased bodies impact loads remain poorly investigated. In this study, tests were conducted on coal, mortar, and specimens using a split Hopkinson pressure bar (SHPB) system. The properties failure behavior loading systematically investigated terms...

10.2113/2022/9211516 article EN cc-by Lithosphere 2022-07-28
Coming Soon ...