Search

Your search keyword '"Feng, Yuchao"' showing total 220 results
Start Over Author "Feng, Yuchao"
220 results on '"Feng, Yuchao"'

1. Transfer Learning in Vocal Education: Technical Evaluation of Limited Samples Describing Mezzo-soprano

Author

Hou, Zhenyi, Zhao, Xu, Ye, Kejie, Sheng, Xinyu, Jiang, Shanggerile, Xia, Jiajing, Zhang, Yitao, Ban, Chenxi, Luo, Daijun, Chen, Jiaxing, Zou, Yan, Feng, Yuchao, Fan, Guangyu, and Yuan, Xin

Subjects
Electrical Engineering and Systems Science - Audio and Speech Processing, Computer Science - Artificial Intelligence, Computer Science - Multimedia, Computer Science - Sound
Abstract

Vocal education in the music field is difficult to quantify due to the individual differences in singers' voices and the different quantitative criteria of singing techniques. Deep learning has great potential to be applied in music education due to its efficiency to handle complex data and perform quantitative analysis. However, accurate evaluations with limited samples over rare vocal types, such as Mezzo-soprano, requires extensive well-annotated data support using deep learning models. In order to attain the objective, we perform transfer learning by employing deep learning models pre-trained on the ImageNet and Urbansound8k datasets for the improvement on the precision of vocal technique evaluation. Furthermore, we tackle the problem of the lack of samples by constructing a dedicated dataset, the Mezzo-soprano Vocal Set (MVS), for vocal technique assessment. Our experimental results indicate that transfer learning increases the overall accuracy (OAcc) of all models by an average of 8.3%, with the highest accuracy at 94.2%. We not only provide a novel approach to evaluating Mezzo-soprano vocal techniques but also introduce a new quantitative assessment method for music education.

Published
2024

2. UP-Diff: Latent Diffusion Model for Remote Sensing Urban Prediction

Author

Wang, Zeyu, Hao, Zecheng, Lin, Jingyu, Feng, Yuchao, and Guo, Yufei

Subjects
Computer Science - Computer Vision and Pattern Recognition, Electrical Engineering and Systems Science - Image and Video Processing
Abstract

This study introduces a novel Remote Sensing (RS) Urban Prediction (UP) task focused on future urban planning, which aims to forecast urban layouts by utilizing information from existing urban layouts and planned change maps. To address the proposed RS UP task, we propose UP-Diff, which leverages a Latent Diffusion Model (LDM) to capture positionaware embeddings of pre-change urban layouts and planned change maps. In specific, the trainable cross-attention layers within UP-Diff's iterative diffusion modules enable the model to dynamically highlight crucial regions for targeted modifications. By utilizing our UP-Diff, designers can effectively refine and adjust future urban city plans by making modifications to the change maps in a dynamic and adaptive manner. Compared with conventional RS Change Detection (CD) methods, the proposed UP-Diff for the RS UP task avoids the requirement of paired prechange and post-change images, which enhances the practical usage in city development. Experimental results on LEVIRCD and SYSU-CD datasets show UP-Diff's ability to accurately predict future urban layouts with high fidelity, demonstrating its potential for urban planning. Code and model weights are available at https://github.com/zeyuwang-zju/UP-Diff., Comment: 5 pages, 4 figures

Published
2024

3. Context-Aware Integration of Language and Visual References for Natural Language Tracking

Author

Shao, Yanyan, He, Shuting, Ye, Qi, Feng, Yuchao, Luo, Wenhan, and Chen, Jiming

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Tracking by natural language specification (TNL) aims to consistently localize a target in a video sequence given a linguistic description in the initial frame. Existing methodologies perform language-based and template-based matching for target reasoning separately and merge the matching results from two sources, which suffer from tracking drift when language and visual templates miss-align with the dynamic target state and ambiguity in the later merging stage. To tackle the issues, we propose a joint multi-modal tracking framework with 1) a prompt modulation module to leverage the complementarity between temporal visual templates and language expressions, enabling precise and context-aware appearance and linguistic cues, and 2) a unified target decoding module to integrate the multi-modal reference cues and executes the integrated queries on the search image to predict the target location in an end-to-end manner directly. This design ensures spatio-temporal consistency by leveraging historical visual information and introduces an integrated solution, generating predictions in a single step. Extensive experiments conducted on TNL2K, OTB-Lang, LaSOT, and RefCOCOg validate the efficacy of our proposed approach. The results demonstrate competitive performance against state-of-the-art methods for both tracking and grounding., Comment: Accepted by CVPR2024

Published
2024

4. Adaptive-Mask Fusion Network for Segmentation of Drivable Road and Negative Obstacle With Untrustworthy Features

Author

Feng, Zhen, Feng, Yuchao, Guo, Yanning, and Sun, Yuxiang

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Robotics
Abstract

Segmentation of drivable roads and negative obstacles is critical to the safe driving of autonomous vehicles. Currently, many multi-modal fusion methods have been proposed to improve segmentation accuracy, such as fusing RGB and depth images. However, we find that when fusing two modals of data with untrustworthy features, the performance of multi-modal networks could be degraded, even lower than those using a single modality. In this paper, the untrustworthy features refer to those extracted from regions (e.g., far objects that are beyond the depth measurement range) with invalid depth data (i.e., 0 pixel value) in depth images. The untrustworthy features can confuse the segmentation results, and hence lead to inferior results. To provide a solution to this issue, we propose the Adaptive-Mask Fusion Network (AMFNet) by introducing adaptive-weight masks in the fusion module to fuse features from RGB and depth images with inconsistency. In addition, we release a large-scale RGB-depth dataset with manually-labeled ground truth based on the NPO dataset for drivable roads and negative obstacles segmentation. Extensive experimental results demonstrate that our network achieves state-of-the-art performance compared with other networks. Our code and dataset are available at: https://github.com/lab-sun/AMFNet., Comment: This paper has been accepted by 2023 IEEE Intelligent Vehicles Symposium (IV) (IEEE IV 2023)

Published
2023

5. GA-HQS: MRI reconstruction via a generically accelerated unfolding approach

Author

Jiang, Jiawei, Feng, Yuchao, Xu, Honghui, Chen, Wanjun, and Zheng, Jianwei

Subjects
Electrical Engineering and Systems Science - Image and Video Processing, Computer Science - Computer Vision and Pattern Recognition
Abstract

Deep unfolding networks (DUNs) are the foremost methods in the realm of compressed sensing MRI, as they can employ learnable networks to facilitate interpretable forward-inference operators. However, several daunting issues still exist, including the heavy dependency on the first-order optimization algorithms, the insufficient information fusion mechanisms, and the limitation of capturing long-range relationships. To address the issues, we propose a Generically Accelerated Half-Quadratic Splitting (GA-HQS) algorithm that incorporates second-order gradient information and pyramid attention modules for the delicate fusion of inputs at the pixel level. Moreover, a multi-scale split transformer is also designed to enhance the global feature representation. Comprehensive experiments demonstrate that our method surpasses previous ones on single-coil MRI acceleration tasks.

Published
2023

6. Formation, Structural Characteristics and Emulsification Properties of Mung Bean Globulin Amyloid Fibrils

Author

BAI Lu, LI Zhiming, ZHANG Shu, FENG Yuchao, FU Tianxin, LIU Yuhang, WANG Changyuan

Subjects
mung bean globulin, protein amyloid fibril, structural characteristics, emulsification properties, Food processing and manufacture, TP368-456
Abstract

In this study, the structural evolution of mung bean globular amyloid fibrils (MBGFs) during the formation process was investigated with different acid-heat treatment times, and the emulsification properties of MBGFs were also explored at each stage of the formation process. The results showed that the subunits of mung bean globulin (MBG) were gradually degraded and hydrolyzed into fibrous structure units mainly composed of small peptides during 0–12 h of heating. And a large number of β-sheet structures were produced, and its relative content increased from an initial level of (18.28 ± 0.75)% to (53.61 ± 1.15)% over the 12 heating period. Enhanced fluorescence intensity was recorded after combination with thioflavin-T. Morphologically, MBGFs gradually became elongated and pliable, and oriented fibril aggregation occurred in the heating process. Between 16 and 24 h, the structure of mature MBGFs gradually was dissociated and the structural characteristics of the fibers were destroyed. The emulsification activity, emulsion stability, protein adsorption rate and interfacial protein content of MBGF were significantly improved compared with those of MBG. MBGFs formed by acid-heat treatment for 4 h had the best emulsification properties. Oil droplets in the MBGF emulsion were small in size, uniformly and orderly distributed, and the emulsion had the highest apparent viscosity and showed an elastic gel structure. In conclusion, different acid-heat treatment times had significant effects on the structure and emulsification properties of MBGFs, and MBGFs had better emulsification properties than MBG. This study will provide theoretical support for clarifying the formation of MBGFs and ideas for the development of highly efficient food-grade emulsifiers.

Published
2024
Full Text
View/download PDF

7. GCN-based Autism Spectrum Disorder Diagnosis via Convolutional Restructuring Attention

Author

Wang, Yibin, Long, Haixia, Zhou, Qianwei, Feng, Yuchao, and Zheng, Jianwei

Subjects
Artificial Intelligence, Computer Science, Neuroscience, Machine learning, Clinical methods, Computational Modeling, Computational neuroscience, fMRI, Neural Networks
Abstract

Brain function connectivity, derived from functional magnetic resonance imaging (fMRI), has enjoyed high popularity in studies of Autism Spectrum Disorder (ASD) diagnosis. Albeit rapid progress has been made, most studies still suffer from several knotty issues: (1) the hardship of effectively modeling the sophisticated brain neuronal connectivity;(2) the dimensionality explosion resulted from excessive voxels in each fMRI sample; (3) the poor interpretability giving rise to unpersuasive diagnosis. To ameliorate these issues, we propose a GCN-based model, namely ConResNet. Specifically, a convolutional restructuring attention is designed for efficiently ROI-interaction feature extraction. Besides, we embed a salient graph pooling method to sift salient nodes. Extensive experiments conducted on ABIDE dataset demonstrate ConResNet achieves state-of-the-art performance. Moreover, the most salient brain regions predicted by ConResNet are closely consistent with theoretical knowledge in the medical domain, providing potential biomarkers for ASD clinical diagnosis.

Published
2023

13. Boosting Feature-Aware Network for Salient Object Detection

Author

Zheng, Jianwei, Gu, Yubin, Feng, Yuchao, Xu, Jinshan, Zhang, Meiyu, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Pimenidis, Elias, editor, Angelov, Plamen, editor, Jayne, Chrisina, editor, Papaleonidas, Antonios, editor, and Aydin, Mehmet, editor

Published
2022
Full Text
View/download PDF

24. Linearly modulated multi-focal diffractive lens for multi-sheet excitation of flow driven samples in a light-sheet fluorescence microscope

Author

Hofmann Meike, Gharbi Ghebjagh Shima, Feng Yuchao, Fan Chao, Lemke Karen, and Sinzinger Stefan

Subjects
light sheet fluorescence microscopy, diffractive optical elements, multi-plane imaging, 3d imaging, multi-focal lens, Applied optics. Photonics, TA1501-1820, Optics. Light, QC350-467
Abstract

Light sheet fluorescence microscope with single light sheet illumination enables rapid 3D imaging of living cells. In this paper we show the design, fabrication and characterization of a diffractive optical element producing several light sheets along a 45° inclined tube. The element, which is based on a multi-focal diffractive lens and a linear grating, generates five thin light sheets with equal intensities when combined with a refractive cylindrical lens. The generated uniform light sheets can be applied for the scanning of samples in tubes enabling flow-driven 3-dimensional imaging.

Published
2023
Full Text
View/download PDF

32. Linearly modulated multi-focal diffractive lens for multi-sheet excitation of flow-driven samples in a light-sheet fluorescence microscope

Author

Hofmann Meike, Gharbi Ghebjagh Shima, Fan Chao, Feng Yuchao, Lemke Karen, and Sinzinger Stefan

Subjects
Physics, QC1-999
Abstract

Light-sheet fluorescence microscopy (LSFM) with single light-sheet illumination enables rapid 3Dimaging of living cells. In this paper we show the design, fabrication and characterization of a diffractive optical element producing several light sheets along an inclined tube for applications in flow-driven imaging. The element, which is based on a multi-focal Fresnel zone plate and a linear grating, generates in combination with a refractive cylindrical lens five thin light sheets of equal intensity.

Published
2022
Full Text
View/download PDF

37. Application and comparison of multiple occupational health risk assessment methods in the footwear industry.

Author

PENG Zhiheng, HE Yinan, FENG Yuchao, ZHANG Hai, DENG Yingcong, ZHOU Liping, ZHOU Hailin, and WANG Zhi

Subjects
HEALTH risk assessment, INDUSTRIAL hygiene, FOOTWEAR industry, MULTIPLE comparisons (Statistics), THRESHOLD limit values (Industrial toxicology)
Abstract

Objective To explore the applicability of different assessment methods in the shoe-making industry, providing a scientific basis for the selection of occupational health risk assessment methods in the shoe -making industry in China. Methods From 2019 to 2020, four shoe -making enterprises in Guangzhou were selected as the study objects. And six methods, including the US Environmental Protection Agency inhalation risk assessment method (US EPA method), the Multiple Exposure Scenarios method (MES method), the International Council on Mining and Metals Occupational Health Assessment Method (ICMM method), the Australian Occupational Health and Safety Risk Assessment Method (Australian method), the GBZ/T 298-2017 "Technical Guidelines for Occupational Health Risk Assessment of Chemical Hazardous Factors in the Workplace" Comprehensive Index Method (Comprehensive Index Method), and the Hazardous Operation Grading Method, were used to conduct occupational health risk assessment for the key positions in the shoe -making enterprises. The risk ratio was used to standardize the risk assessment results derived from these six methods, and the results of the six methods were compared. Results The airborne concentrations of benzene, 1,2 -dichloroethane, trichloroethylene, and n-Hexane in the key positions of these four shoe -making enterprises were lower than the national occupational exposure limit. There were differences in results derived from six methods. The pairwise comparison showed that the risk levels assessed by the US EPA method, the Comprehensive Index Method, the MES method, the ICMM method, and the Australian method were statistically higher than those assessed with the Hazardous Operation Grading Method (P < 0.05). Among them, the risk level assessed by the US EPA method was the highest, followed by the Comprehensive Index Method; similar risk levels were derived by the MES method and the Australian method; and the ICMM method had a large difference from other assessment methods. Conclusions It is recommended to use the Comprehensive Index Method as the main method, with both the MES method and the Australian method as the supplementary methods. A combination of multiple assessment methods is suggested to conduct occupational health risk assessment in the shoe-making industry. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

42. IFI-16 Recognizing Leakage Mitochondrial DNA Inducing Inflammation in Myocardial Remodeling Post Infarction Through Inflammasome and Interacts with Galectin-3

Author

Deng, Yi, primary, Feng, Yuchao, additional, Pang, Xiuqing, additional, Huang, Xiaoyan, additional, Peng, Weihang, additional, Huang, Peiying, additional, Zhao, Shuai, additional, Li, Zhishang, additional, Chen, Li, additional, Cai, Xingui, additional, Huang, Qiuping, additional, Zeng, Jing, additional, Li, Qingmin, additional, and Chen, Bojun, additional

Published
2023
Full Text
View/download PDF

46. Fast Tensor Nuclear Norm for Structured Low-Rank Visual Inpainting

Author

Jianwei Zheng, Feng Yuchao, Xu Honghui, Xiaomin Yao, and Shengyong Chen

Subjects
Similarity (geometry), Rank (linear algebra), Computer science, Property (programming), Singular value decomposition, Media Technology, Matrix norm, Inpainting, Tensor, Electrical and Electronic Engineering, Algorithm, Image (mathematics)
Abstract

Low-rank modeling has achieved great success in visual data completion. However, the low-rank assumption of original visual data may be in approximate mode, which leads to suboptimality for the recovery of underlying details, especially when the missing rate is extremely high. In this paper, we go further by providing a detailed analysis about the rank distributions in Hankel structured and clustered cases, and figure out both non-local similarity and patch-based structuralization play a positive role. This motivates us to develop a new Hankel low-rank tensor recovery method that is competent to truthfully capture the underlying details with sacrifice of slightly more computational burden. First, benefiting from the correlation of different spectral bands and the smoothness of local spatial neighborhood, we divide the visual data into overlapping 3D patches and group the similar ones into individual clusters exploring the non-local similarity. Second, the 3D patches are individually mapped to the structured Hankel tensors for better revealing low-rank property of the image. Finally, we solve the tensor completion model via the well-known alternating direction method of multiplier (ADMM) optimization algorithm. Due to the fact that size expansion happens inevitably in Hankelization operation, we further propose a fast randomized skinny tensor singular value decomposition (rst-SVD) to accelerate the periteration running efficiency. Extensive experimental results on real world datasets verify the superiority of our method compared to the state-of-the-art visual inpainting approaches.

Published
2022
Full Text
View/download PDF

50. Chaihu Shugan powder alleviates liver inflammation and hepatic steatosis in NAFLD mice: A network pharmacology study and in vivo experimental validation

Author

Lei, Sisi, Zhao, Shuai, Huang, Xiaoyan, Feng, Yuchao, Li, Zhishang, Chen, Li, Huang, Peiying, Guan, Hansu, Zhang, Haobo, Wu, Qihua, and Chen, Bojun

Subjects
Pharmacology, Pharmacology (medical)
Abstract

Background: Nonalcoholic fatty liver disease (NAFLD) is the most common metabolic disease and is intertwined with cardiovascular disorders and diabetes. Chaihu Shugan powder (CSP) is a traditional Chinese medicine with a significant therapeutic effect on metabolic diseases, such as NAFLD. However, its pharmacological mechanisms remain to be elucidated.Methods: The main compounds of CSP were measured using LC-MS/MS. A network pharmacology study was conducted on CSP. Its potential active ingredients were selected according to oral bioavailability, drug similarity indices, and phytochemical analysis. After obtaining the intersected genes between drug targets and disease-related targets, the component-disease-target network and protein-protein interaction analysis were visualized in Cytoscape. GO and KEGG enrichment analyses were performed using the Metascape database. Six-week-old male C57BL/6 mice fed a high-fat high-fructose diet for 16 weeks plus chronic immobilization stress for 2 weeks, an in vivo model, were administered CSP or saline intragastrically. Liver histology, triglyceride and cholesterol levels, ELISA, and RT-PCR were used to assess hepatic inflammation and steatosis. Immunohistochemistry and western blotting were performed to assess protein levels.Results: A total of 130 potential target genes in CSP that act on NAFLD were identified through network pharmacology assays, including tumor necrosis factor (TNF), interleukin-6 (IL6), interleukin-1β (IL-1β), and peroxisome proliferator-activated receptor γ (PPARG). KEGG enrichment analysis showed that the main pathways were involved in inflammatory pathways, such as the TNF and NF-κB signaling pathways, and metabolism-related pathways, such as the MAPK, HIF-1, FoxO, and AMPK signaling pathways. The results in vivo showed that CSP ameliorated liver inflammation and inhibited hepatic fatty acid synthesis in the hepatocyte steatosis model. More specifically, CSP therapy significantly inhibited the expression of tumor necrosis factor α (TNFα), accompanied by a decrease in TNF receptor 1 (TNFR1) and the ligand availability of TNFR1.Conclusion: Through the combination of network pharmacology and in vivo validation, this study elucidated the therapeutic effect of CSP on NAFLD, decreasing liver inflammation and inhibiting hepatic fatty acid synthesis. More specifically, the anti-inflammatory action of CSP was at least partially mediated by inhibiting the TNFα/TNFR1 signaling pathway.

Published
2022
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results