检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Ying Hu Jiabo Jing Fan Li Lijun He Li Lin Wenzhong Yang Key Laboratory of Signal Detection and Processing Urumqi China School of Computer Science and Technology Xinjiang University Urumqi China School of Information and Communications Engineering Xi’an Jiaotong University Xi’an China Xinjiang Institute of Electronics Research Shares Co. Ltd Urumgi China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Extracting singing melody from polyphonic music is an important topic in the field of music information retrieval. In this paper, we propose a singing melody extraction network consisting of five stacked multi-scale feature time-frequency aggregation (MF-TFA) modules. In the same network, deeper layers generally contain more contextual information than shallower layers. To help the shallower layers enhance the ability of task-relevant feature extraction, we propose a self-distillation and multi-level supervision (SD-MS) method, which leverages the feature distillation from the deepest layer to the shallower one and multi-level supervision to guide network training. Visualization analysis shows that by introducing SD-MS, the same-level layer in the network can obtain a clearer representation of fundamental frequency components, while the shallower layers can even learn more task-relevant semantic information. Ablation study results indicate that SD-MS applies to existing melody extraction models and can consistently improve performance. Experimental results show that our proposed method, MF-TFA with SD-MS, outperforms six compared state-of-the-art methods, achieving overall accuracy (OA) scores of 87.1%, 89.9%, and 76.6% on the ADC 2004, MIREX 05, and MEDLEY DB datasets, respectively. The main code will be available at https://***/SmoothJing/MF-TFA_SD-MS.

关键词： Training Visualization Time-frequency analysis Semantics signal processing Feature extraction Multiple signal classification Data mining Speech processing Music information retrieval

来源：评论

学校读者我要写书评

暂无评论

MN-Net: Speech Enhancement Network via Modeling the Noise

IEEE Transactions on Audio, Speech and Language Processing

引用

IEEE Transactions on Audio, Speech and Language processing 2025年 33卷 1208-1219页

作者： Ying Hu Qin Yang Wenbing Wei Li Lin Liang He Zhijian Ou Wenzhong Yang School of Computer Science and Technology Key Laboratory of Signal Detection and Processing Xinjiang University Urumqi China Xinjiang Institute of Electronics Research Shares Company Ltd. Urumqi China Department of Electrical Engineering Tsinghua University Beijing China Beijing National Research Center for information Science and Technology Department of Electronics Tsinghua University Beijing China

Currently, deep learning-based speech enhancement methods generally focus on target speech extraction while neglecting modeling the other sound sources in the mixture. These methods still can't distinguish the target speech from the interference well. In this paper, we present a monaural speech enhancement network via Modeling the Noise (MN-Net), which includes a shared Encoder and three separate Decoders for parallel modeling the magnitude and phase spectrogram of target speech, and the complex spectrogram of noise. Specifically, we propose a Multi-Branch Feature Extractor (MBFE) module to capture the richer contextual information in mixture, and a Spatial Reconstruction Unit (SRU) to remove the redundancy from extracted features. We compared our proposed MN-Net with 18 classical speech enhancement methods on the VoiceBank+DEMAND dataset, and with 9 ones on DNS-Challenge dataset for denoising task, and with 7 ones on the WHAMR! dataset for simultaneous denoising & de-reverberation task. Our proposed MBFE module was applied to two classical speech enhancement methods, DB-AIAT and CMGAN, replacing their DenseBlocks module. The results demonstrate that applying the MBFE module can boost their performances while keeping smaller model size. A series of visualization analysis intuitively verify that modeling the noise can enable the network to distinguish the target speech from noise and other interference more accurately.

关键词： Noise Feature extraction Spectrogram Speech enhancement Decoding Kernel Transformers Noise reduction Data mining Training

来源：评论

学校读者我要写书评

暂无评论

Stable Extended U-Net for Noise-Robust Speaker Verification

Stable Extended U-Net for Noise-Robust Speaker Verification

引用

International Conference on Acoustics, Speech, and signal processing (ICASSP)

作者： Zonghui Wang Zhihua Fang Liang He School of Computer Science and Technology Xinjiang University Urumqi China Xinjiang Key Laboratory of Signal Detection and Processing Urumqi China School of Intelligence Science and Technology Xinjiang University Urumqi China Department of Electronic Engineering Tsinghua University Beijing China Beijing National Research Center for Information Science and Technology Tsinghua University Beijing China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

With advancements in deep learning, speaker verification systems have significantly improved their performance in noisy environments. Researchers typically demonstrate the effectiveness of their improved models by comparing performance on specific datasets, such as the VoxCeleb benchmark. However, in diverse real-world noise conditions, the out-of-domain generalization ability is also a crucial factor in evaluating a model’s performance improvement. Research on stable learning indicates that eliminating the spurious correlation between training and testing data can enhance the generalization of the model. Building on this idea, we propose an improved speaker verification system with high generalization based on the extended U-Net (ExU-Net). It uses the sample reweighting method from stable learning to eliminate sample correlations and retains more effective speaker information through subpixel convolutions and coordinate attention mechanisms. We validate the effectiveness of this approach through extensive evaluations on VoxCeleb1, VOiCES, and other out-of-domain noise test sets, highlighting its generalization capability and model robustness.

关键词： Training Correlation Attention mechanisms Convolution Data models Stability analysis Robustness Noise measurement Speech processing signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

GCW-YOLOv8n: Lightweight Safety Helmet Wearing detection Algorithm

引用

Computer Engineering and Applications 2025年第3期61卷 144-154页

作者： Xu, Zhuang Qian, Yurong Yan, Feng Key Laboratory of Software Engineering Xinjiang University Urumqi830091 China Key Laboratory of Signal Detection and Processing in Xinjiang Uygur Autonomous Region Xinjiang University Urumqi830046 China School of Computer Science and Technology Xinjiang University Urumqi830046 China

China is a major industrial country in the world. In various construction environments, the falling of construction materials and collisions on construction sites are the main causes of casualties. Accidents caused by head injuries often occur, and wearing safety helmets can ensure the safety of construction personnel to the greatest extent possible. In order to solve the problems of poor timeliness and low management efficiency of manual management, the existing models have strict requirements for computing power, large memory requirements, and handling of load and data transmission delay of industrial equipment, and to achieve edge computing and real-time control, a modified helmet wearing detection algorithm based on YOLOv8n is proposed. Firstly, a new GS-C2f module is proposed, which introduces GhostConv and SE (squeeze and excitation) attention mechanism, effectively reducing the computational complexity of the model and helping the network extract features effectively. Secondly, the CBAM attention mechanism is introduced in the Neck section to enhance the model focus on effective features. Finally, Wise-IoUv3 is introduced to further improve the accuracy of the model. Through experiments, compared with the original YOLOv8n model, this model achieves a 21.24% reduction in computational parameters and a 0.01 improvement in recognition accuracy, achieving satisfactory results between model accuracy and complexity. © 2025 Journal of Computer Engineering and Applications Beijing Co., Ltd.;Science Press. All rights reserved.

关键词： Office equipment

来源：评论

学校读者我要写书评

暂无评论

Feature-Consistent Cyclic Domain Adversarial Network for Unsupervised Transfer Diagnosis of the Autoimmune Disease Igg4-Rd by Raman Spectroscopy

SSRN

引用

SSRN 2025年

作者： Zhang, Ziyang Chen, Cheng Qin, Wentao Wu, Xue Liu, Yang Wu, Lijun Lv, Xiaoyi Chang, Chenjie Chen, Xinya Chen, Zhiyuan Xu, Qianxi Li, Zhuoyu Ma, Mengsi Chen, Chen Xinjiang University College of Software Urumqi830046 China Xinjiang Cloud Computing Application Laboratory Xinjiang University Karamay834099 China School of Computer Science and Technology Xinjiang University Urumqi830046 China Key Laboratory of signal detection and processing Xinjiang University Urumqi830046 China Department of Rheumatology＆Immunology People's Hospital of Xinjiang Uygur Autonomous Region Xinjiang Urumqi China Xinjiang Clinical Research Center for Rheumatoid arthritis Xinjiang Urumqi China

Abstracts IgG4-related disease (IgG4-RD) is a multi-organ immune disorder characterized by systemic involvement, diverse pathogenesis, and rarity, which complicates its diagnosis. Traditional medical diagnostic models, predominantly based on supervised learning in machine learning or deep learning frameworks, struggle to address the scarcity of labeled data due to the diagnostic challenges of IgG4-RD. To overcome this limitation, this study proposes a novel framework based on Feature-Consistent Cyclic Domain Adversarial Networks (FCCDAN). The framework first trains a classification model on the source domain for diagnosing source domain diseases under supervised conditions. Subsequently, leveraging the source domain model, the encoder is guided to extract semantic features, while feature consistency loss is integrated into both the generator and discriminator to align the semantic features of the target domain with those of the source domain. Finally, the aligned features are processed through a pre-trained linear mapping to achieve unsupervised classification and diagnosis in the target domain. The proposed model utilizes common autoimmune diseases as a knowledge source, enabling unsupervised transfer diagnosis for IgG4-RD, and demonstrates generalizability through experiments on other disease datasets, achieving an average diagnostic accuracy of 90.62%. This represents improvements of 4.14%, 9.82%, and 10.51% over baseline models, respectively. FCCDAN expands the applicability of Raman spectroscopy in disease diagnosis, maintaining high performance even in label-scarce environments. The overall research workflow also establishes a robust paradigm for the application and advancement of domain adaptation techniques in this context. © 2025, The Authors. All rights reserved.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

A Lightweight Speech Emotion Recognition Model with Bias-Focal Loss 6

A Lightweight Speech Emotion Recognition Model with Bias-Foc...

引用

6th International Conference on Information Communication and signal processing, ICICSP 2023

作者： Hou, Shijing Yang, Huamin Hu, Ying Xinjiang University Key Laboratory of Signal Detection and Processing Urumqi China

ISBN: (纸本)9798350339994

A major challenge in speech emotion recognition (SER) is how to build a lightweight model with limited training data for applying on the devices with limited in resources. In this paper, we propose a lightweight speech emotion recognition (SER) model with Bias-Focal loss function, where a Dynamic Separable Convolution (DySC) block is designed to extract more fine-grained emotional features and makes the model smaller. We propose a Bias-Focal loss to address the issue of the inconsistent of training samples, while focusing on data points with high feature diversity during the training phase. Experimental results show that our proposed lightweight model is the smallest comparing with other methods, while the number of parameters in our proposed model is 0.85M. Meanwhile, our proposed model achieves the best performance that the score of Unweighted Accuracy (UA) is 75.01 %, and that of Weighted Accuracy (WA) and F1-score are 74.05% and 74.29 % on the IEMOCAP (scripted+improvised)dataset. © 2023 IEEE.

关键词： Bias-Focal loss dynamic separable convolution lightweight Speech emotion recognition

来源：评论

学校读者我要写书评

暂无评论

Research on Shaking Video Image Stabilization Algorithm for Moving Object detection 2

Research on Shaking Video Image Stabilization Algorithm for ...

引用

2nd International Conference on Algorithm, Image processing and Machine Vision, AIPMV 2024

作者： Wang, Chuangchuang Jia, Zhenhong Huang, Xiaohui Song, Sensen Wang, Jiajia Zhou, Gang Shi, Fei Xinjiang University Key Laboratory of Signal Detection and Processing Urumqi China

ISBN: (纸本)9798350390254

Video captured by surveillance equipment will jitter due to the shaking of the equipment, this jitter will affect the detection results of moving target detection algorithms that rely on stable video frames. This paper proposes an improved ORB algorithm to solve the video jitter problem, so that the moving target detection algorithm can accurately detect moving targets in jittery videos. First, wavelet transform is used to locate the high-frequency area of the image, and then feature extraction is performed on the area, which improves the efficiency of ORB feature extraction. And use Boosted Efficient Binary Local Image Descriptor (BEBLID) to replace the descriptors of directional FAST and rotation BRIEF (ORB) to improve matching accuracy. In the feature matching stage, a neighborhood query method is proposed to replace the global search of the traditional BFMatcher, which improves the matching speed. Finally, Progressive Sample Consistency (PROSAC) is employed to ensure accurate matching of point pairs, resulting in a motion matrix for video stabilization. Finally, a Gaussian mixture model with adaptive distribution numbers is combined to quickly detect moving targets. Comparative experiments with scale-invariant feature transform(SIFT), Accelerated-KAZE(AKAZE), ORB, Qtree _ ORB and SIRB prove the superior accuracy and speed of this algorithm. © 2024 IEEE.

关键词： Wavelet transforms

来源：评论

学校读者我要写书评

暂无评论

Improved Self-Consistency Training with Selective Feature Fusion for Sound Event detection 6

Improved Self-Consistency Training with Selective Feature Fu...

引用

6th International Conference on Information Communication and signal processing, ICICSP 2023

作者： Wang, Mingyu Li, Yunlong Hu, Ying Xinjiang University Key Laboratory of Signal Detection and Processing Urumqi China

ISBN: (纸本)9798350339994

Sound event detection (SED) is a joint task of identifying the categories and time boundaries of sound events within an audio clip. In this paper, we propose an improved self-consistency training (ISCT) strategy for semi-supervised SED based on Mean Teacher (MT) method. For teacher and student models, each adopts two branches with the same CRNN structure, the two branches help training the model by means of consistency regularization. ISCT strategy incorporates self-consistency loss on the basis of MT loss to improve the generalization performance of the model. A selective feature fusion (SFF) module is designed for applying in the shallow layers of the feature extraction part to selectively fuse the features with different scales. A parallel attention (PA) module is designed for applying in the deep layers of the feature extraction part to obtain much richer high-level features by the channel and spatial-wise attention. Ablation experiments verify the effectiveness of our proposed ISCT strategy, SFF and PA modules. In addition, compared with four methods, our proposed method achieves competitive performance on the DCASE 2020 task4 dataset. © 2023 IEEE.

关键词： attention selective feature fusion self-consistency training sound event detection

来源：评论

学校读者我要写书评

暂无评论

MCFTNet: Multimodal Cross-layer Fusion Transformer Network for Hyperspectral and LiDAR Data Classification

引用

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2025年

作者： Huang, Wei Wu, Tianren Zhang, Xueyu Li, Liangliang Lv, Ming Jia, Zhenhong Zhao, Xiaobin Ma, Hongbing Vivone, Gemine Xinjiang University School of Computer Science and Technology Urumqi830017 China Xinjiang University Key Laboratory of Signal Detection and Processing Urumqi830017 China Guangxi University School of Computer and Electronic Information Nanning530004 China Guangxi University Guangxi Key Laboratory of Multimedia Communications and Network Technology Nanning530004 China Beijing Institute of Technology School of Information and Electronics Beijing100081 China University of Science and Technology Beijing School of Computer and Communication Engineering Beijing100083 China Tsinghua University Department of Electronic Engineering Beijing100084 China National Research Council Tito85050 Italy NBFC National Biodiversity Future Center Palermo90133 Italy

Remote sensing image classification is a popular yet challenging field. Many researchers have combined convolutional neural networks (CNNs) and Transformers for hyperspectral image (HSI) classification tasks. However, in traditional Transformers, shallow-level information does not propagate well to deeper layers, which can lead to spatial variations and overfitting. Moreover, traditional Transformer models use an external classification (CLS) token that is randomly initialized and often struggles to generalize effectively. In this paper, we combine the strengths of HSI and light detection and ranging (LiDAR) data, using LiDAR as an external CLS token, which significantly enhances classification accuracy and reliability. We propose a new multimodal cross-layer fusion transformer network (MCFTNet), integrating CNNs with the latest Transformer networks. It includes a CNN for extracting spatial features and a hybrid cross-patch attention mechanism (HyCrossAttention) for land cover classification, leveraging LiDAR data to generate CLS and HSI patch tokens. More importantly, to reduce the loss of valuable information during the layer-by-layer propagation, we designed cross-layer skip connections. Through adaptive learning, these cross-layer fusions help address the gradient vanishing problem in deep networks while preserving early-layer features. This enables the model to better integrate information from different layers, enhancing both its stability and performance. We carried out in - depth experiments on commonly - used benchmark datasets, specifically, the University of Houston dataset, the Trento dataset, and the University of Southern Mississippi Gulf Park (MUUFL) dataset. We compared the results of the proposed MCFTNet model with those obtained from state-of-the-art Transformer models, classical CNNs, and traditional classifiers. As a result, the MCFTNet model outshone them all in terms of performance. © 2008-2012 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Localized Target detection Algorithms in Low Illumination Wide-Field of View High-Resolution Environments 24

Localized Target Detection Algorithms in Low Illumination Wi...

引用

4th International Conference on Artificial Intelligence, Big Data and Algorithms, CAIBDA 2024

作者： Xie, Chengyou Huang, Xiaohui Jia, Zhenhong Song, Sensen Wang, Jiajia Zhou, Gang Shi, Fei Key Laboratory of Signal Detection and Processing Xinjiang University Urumqi Xinjiang China

ISBN: (纸本)9798400710247

To solve the problem of difficult detection of small targets in wide-field videos captured by eagle-eye equipment in low-illumination environments. we propose a local moving target detection algorithm based on sample consistency. First, the initial background model is established using the first ten frames of the video. The contrast and texture features between the current and background frames are calculated, and the S component in the HSV space is extracted and fused to obtain the final features. Finally, The target area is located through the global segmentation map and the local tracking detection module is activated. Experimental results demonstrate that Compared to other algorithms, the proposed algorithm reduces the running time by implementing local detection. It effectively minimizes the interference caused by light sources and dark spot noise in low-light environments, thereby enhancing the accuracy of the algorithm. © 2024 Copyright held by the owner/author(s).

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：