检索结果-内蒙古大学图书馆

Complex Neural Spatial Filter: Enhancing Multi-Channel Target Speech Separation in Complex Domain

IEEE SIGNAL processing LETTERS 2021年 28卷 1370-1374页

作者： Gu, Rongzhi Zhang, Shi-Xiong Zou, Yuexian Yu, Dong Peking Univ Shenzhen Grad Sch Sch Elect & Comp Sci ADSPLab Shenzhen 518055 Peoples R China Tencent AI Lab Bellevue WA 98004 USA

To date, mainstream target speech separation (TSS) approaches are formulated to estimate the complex ratio mask (cRM) of target speech in time-frequency domain under supervised deep learning framework. However, the existing methods are designed in the way that the real and imaginary parts of the cRM are separately modeled using real-valued training data pairs. The research motivation of this study is to design a deep model that fully exploits the temporal-spectral-spatial information of multi-channel signals for estimating cRM directly and efficiently in complex domain. As a result, a novel TSS network is designed consisting of two modules, a complex neural spatial filter (cNSF) and an MVDR. Essentially, cNSF is a cRM estimation model and an MVDR module is cascaded to the cNSF module to reduce the nonlinear speech distortions introduced by neural network. Specifically, to fit the cRM target, all input features of cNSF are reformulated into complex-valued representations. Then, to achieve good hierarchical feature abstraction, a complex deep neural network (cDNN) is delicately designed with U-Net structure. Experiments conducted on simulated multi-channel speech data demonstrate the proposed cNSF outperforms the baseline NSF by 12.1% scale-invariant signal-to-distortion ratio and 33.1% word error rate.

关键词： Customer relationship management Spectrogram Training Task analysis Supervised learning Speech enhancement Nonlinear distortion Target speech separation complex deep neural networks complex ratio mask estimation MVDR

来源：评论

学校读者我要写书评

暂无评论

Analysis of Swin-UNet vision transformer for Inferior Vena Cava filter segmentation from CT scans ☆

引用

ARTIFICIAL INTELLIGENCE IN THE LIFE SCIENCES 2023年 4卷

作者： Gomes, Rahul Pham, Tyler He, Nichol Kamrowski, Connor Wildenberg, Joseph Univ Wisconsin Eau Claire Dept Comp Sci Eau Claire WI 54701 USA Univ Minnesota Twin Cities Dept Comp Sci Minneapolis MN 55455 USA Mayo Clin Hlth Syst Intervent Radiol Eau Claire WI 54703 USA

Purpose: The purpose of this study is to develop an accurate deep learning model capable of Inferior Vena Cava (IVC) filter segmentation from CT scans. The study does a comparative assessment of the impact of Residual Networks (ResNets) complemented with reduced convolutional layer depth and also analyzes the impact of using vision transformer architectures without performance degradation. Materials and Methods: This experimental retrospective study on 84 CT scans consisting of 54618 slices involves design, implementation, and evaluation of segmentation algorithm which can be used to generate a clinical report for the presence of IVC filters on abdominal CT scans performed for any reason. Several variants of patch-based 3D-Convolutional Neural Network (CNN) and the Swin UNet Transformer (Swin-UNETR) are used to retrieve the signature of IVC filters. The Dice Score is used as a metric to compare the performance of the segmentation models. Results: Model trained on UNet variant using four ResNet layers showed a higher segmentation performance achieving median Dice = 0.92 [Interquartile range(IQR): 0.85, 0.93] compared to the plain UNet model with four layers having median Dice = 0.89 [IQR: 0.83, 0.92]. Segmentation results from ResNet with two layers achieved a median Dice = 0.93 [IQR: 0.87, 0.94] which was higher than the plain UNet model with two layers at median Dice = 0.87 [IQR: 0.77, 0.90]. Models trained using SWIN-based transformers performed significantly better in both training and validation datasets compared to the four CNN variants. The validation median Dice was highest in 4 layer Swin UNETR at 0.88 followed by 2 layer Swin UNETR at 0.85. Conclusion: Utilization of vision based transformer Swin-UNETR results in segmentation output with both low bias and variance thereby solving a real-world problem within healthcare for advanced Artificial Intelligence (AI) image processing and recognition. The Swin UNETR will reduce the time spent manually tracking IVC f

关键词： deep learning Medical imaging Convolutional neural networks SWIN transformer UNet ResNet IVC filter

来源：评论

学校读者我要写书评

暂无评论

TOWARDS EFFICIENT MODELS FOR real-time deep NOISE SUPPRESSION

TOWARDS EFFICIENT MODELS FOR REAL-TIME DEEP NOISE SUPPRESSIO...

引用

IEEE International Conference on Acoustics, Speech and Signal processing (ICASSP)

作者： Braun, Sebastian Gamper, Hannes Reddy, Chandan K. A. Tashev, Ivan Microsoft Res Redmond WA 98052 USA

ISBN: (纸本)9781728176055

With recent research advancements, deep learning models are becoming attractive and powerful choices for speech enhancement in real-time applications. While state-of-the-art models can achieve outstanding results in terms of speech quality and background noise reduction, the main challenge is to obtain compact enough models, which are resource efficient during inference time. An important but often neglected aspect for data-driven methods is that results can be only convincing when tested on real-world data and evaluated with useful metrics. In this work, we investigate reasonably small recurrent and convolutional-recurrent network architectures for speech enhancement, trained on a large dataset considering also reverberation. We show interesting tradeoffs between computational complexity and the achievable speech quality, measured on real recordings using a highly accurate MOS estimator. It is shown that the achievable speech quality is a function of network complexity, and show which models have better tradeoffs.

关键词： speech enhancement noise reduction convolutional recurrent neural network efficient neural networks

来源：评论

学校读者我要写书评

暂无评论

A Vision Transformer Approach to Efficient Brain Tumor Detection and Segmentation in Medical Imaging

A Vision Transformer Approach to Efficient Brain Tumor Detec...

引用

Data Science, Agents & Artificial Intelligence (ICDSAAI), International Conference on

作者： Ebinesh K Aashish D Balaji Sougoumar Anand Karuppannan B.E. Electronics Engineering (VLSI Design and Technology) Chennai Institute of Technology Chennai India Department of Computer Science and Engineering Chennai Institute of Technology Chennai India Department of Electronics and Communication Engineering Chennai Institute of Technology Chennai India

ISBN: (数字)9798331537555

ISBN: (纸本)9798331537562

Brain tumor detection and segmentation are important tasks in the field of medical imaging that have significant potential to enhance treatment planning and accurate diagnosis. The construction of a Vision Transformer (ViT) model for the purpose of identifying brain tumor from medical images specifically MRI scans has been covered in this work. Renowned for its outstanding performance recognition challenges, the Vision Transformer will be utilized to improve brain tumor detection accuracy and efficiency. The proposed system will be evaluated in terms of accuracy, performance, and inference time against conventional deep learning models. This research aims to improve real-time medical image processing with ViT, resulting in faster diagnosis and more efficient tumor identification in clinical practice.

关键词： Computer vision image segmentation Accuracy Hospitals Magnetic resonance imaging Brain tumors Transformers Brain modeling real-time systems Planning

来源：评论

学校读者我要写书评

暂无评论

Support Vector Machines: Unveiling the Power and Versatility of SVMs in Modern Machine learning 3

Support Vector Machines: Unveiling the Power and Versatility...

引用

3rd International Conference on Innovative Mechanisms for Industry Applications, ICIMIA 2023

作者： Saravanan, K. Prakash, R.Banu Balakrishnan, C. Kumar, Gade Venkata Prasanna Siva Subramanian, R. Anita, M. R.M.D. Engineering College Department of Information Technology Kavaraipettai601206 India Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Dept of CSE India S.A Engineering College Dept of CSE India Engineering Insight Software International PVT Ltd India R.M.K College of Engineering and Technology Dept of CSE India Easwari Engineering College Dept of IT India

ISBN: (纸本)9798350343632

Support vector machines, or SVMs, have become a really big deal in machine learning because of how good they are at classification and regression problems. This article explores in-depth knowledge about SVMs in ML algorithms. First from the history of SVMs, starting with when they were first thought up and addresses some important stuff different researchers have done with them over time. Next, get into the math and theory behind how SVMs work things like margins support vectors, and optimization problems and also discuss different ways SVMs have been tweaked and changed, like versions for multiple classes, support vector regression one-class SVMs, and twin SVMs. Another key part of SVMs is kernel functions here we spend a bunch of time breaking those down and explaining what they do to transform data so SVMs can work with it better. Further on, we look at some real-world uses for SVMs, like in image recognition natural language processing, bioinformatics, and finance. Even though SVMs have been around for a while they're still super relevant today, so here summarize the most important discoveries about them and think about ways SVMs might keep evolving or being used differently as machine learning keeps moving fast. Overall, the goal here is to give you a deep dive into Support Vector Machines - where they began, the technical details, how they are applied, and where they might go in the future helpful in better understanding the SVM in ML Algorithm. © 2023 IEEE.

关键词： Kernal Functions Linear SVM Machine learning Optimization SVM

来源：评论

学校读者我要写书评

暂无评论

Nighttime object detection system with lightweight deep network for internet of vehicles

引用

JOURNAL OF real-time image processing 2021年第4期18卷 1141-1155页

作者： Jhong, Sin-Ye Chen, Yung-Yao Hsia, Chih-Hsien Lin, Shih-Chang Hsu, Kuo-Hua Lai, Chin-Feng Natl Cheng Kung Univ Dept Engn Sci Tainan Taiwan Natl Taiwan Univ Sci & Technol Dept Elect & Comp Engn Taipei Taiwan Natl Taipei Univ Technol Grad Inst Automat Technol Taipei Taiwan

Autonomous driving systems in internet of vehicles (IoV) applications usually adopt a cloud computing mode. In these systems, information got at the edge of the cloud computing center for data analysis and situation response. However, the conventional IoV face enormous challenges to meet the requirements in terms of storage, communication, and computing problems because of the considerable amount of information on the traffic environment. The environment perception during the nighttime is poorer than that during the daytime that this problem also requires addressing. To solve these problems, we propose a nighttime object detection scheme based on a lightweight deep learning model in the edge computing mode. First, the pedestrian detection and the vehicle detection algorithm that using the thermal images based on the YOLO architecture. We can implement the model on edge devices that can achieve real-time detection through the designed lightweight strategy. Next, a spatial prior information and temporal prior information into the detection algorithm and divide the frames into key and non-key frames to increase the performance and speed of the system simultaneously. Finally, we implemented the detection network for performance and feasibility verification on the Jetson TX2 edge device. The experimental results show that the proposed system can achieve real-time and high-accuracy object detection on edge devices.

关键词： Internet of vehicles Thermal infrared images Object detection deep neural network Prior information

来源：评论

学校读者我要写书评

暂无评论

Flower Phenotype Recognition and Analysis using YoloV5 Models 13

Flower Phenotype Recognition and Analysis using YoloV5 Model...

引用

13th International Conference on Advances in Computing, Control, and Telecommunication Technologies, ACT 2022

作者： Naik, Praveen M. Rudra, Bhawana Department of Information Technology National Institute of Technology Karnataka Surathkal Mangaluru575025 India

ISBN: (纸本)9781713857938

real time detection of flowers based on their growth cycle is called as phenotype. It is one of the most important methods for judging the maturity of flowers and to estimate their yield. Traditional method involves flower detection and classification of varied species. In this paper, we introduce a new dataset based on flower phenotype. The dataset has images of flowers, classified into bud, fresh and stale. The work will help for identification and localization of classes based on flower phenotype. The detection of flowers at various stages of their life will be more important to harvesting in floriculture field. We propose a state of art deep learning-based approach using YOLOV5 model is used for identifying flowers based on flower phenotype. The images are subsequently augmented using rotation transformation, color balance transformation, brightness transformation and blur processing. The augmented images are used for preparing training sets. Using different versions of YoloV5 the flower image dataset is trained and tested. Primarily, flower phenotype considered has stages like bud formation, blossoming and stale flower. © Grenze Scientific Society, 2022.

关键词： deep learning

来源：评论

学校读者我要写书评

暂无评论

A Survey of Track Prediction Method of AUV Based on deep learning

A Survey of Track Prediction Method of AUV Based on Deep Lea...

引用

International Conference on image, Vision and Intelligent Systems, ICIVIS 2021

作者： Yu, Yuna Zhang, Jing Zhang, Tianchi School of Information Science and Engineering University of Jinan Jinan250022 China Shandong Provincial Key Laboratory of Network-Based Intelligent Computing Jinan250022 China School of Information Science and Engineer Chongqing Jiaotong University Chongqing400074 China

ISBN: (纸本)9789811669620

Autonomous Underwater Vehicles (AUVs) work in the complex marine environment, and security is the focus of research and application. Track prediction is the key to assist AUV to avoid risk, and it is of great significance to predict the drift track of the AUV in the accident for the later study of fault causes and solutions. The real-time spatial position and ocean environment of AUV are the basic data of track prediction, which are large spatial data with high real-time performance. In this paper, the establishment and development of AUV track prediction model are analyzed based on the characteristics of deep learning that can efficiently process complex data. Secondly, the drift track prediction of AUV with large diving depth after the accident is introduced with the focus of the drift track prediction after the AUV floats to the sea. Finally, the research on the track drifting with the ocean current is prospected to face the situation when AUV floating slowly under the effect of residual buoyancy after the accident of AUV. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

A universal method for solar filament detection from Hα observations using semi-supervised deep learning

引用

ASTRONOMY & ASTROPHYSICS 2024年第/06期686卷

作者： Diercke, Andrea Jarolim, Robert Kuckein, Christoph Manrique, Sergio J. Gonzalez Ziener, Marco Veronig, Astrid M. Denker, Carsten Poetzi, Werner Podladchikova, Tatiana Pevtsov, Alexei A. Inst Sonnenphys KIS Georges Kohler Allee 401 D-79110 Freiburg Germany Natl Solar Observ NSO 3665 Discovery Dr Boulder CO 80303 USA Leibniz Inst Astrophys Potsdam AIP Sternwarte 16 D-14482 Potsdam Germany Univ Potsdam Inst Phys & Astron Karl Liebknecht Str 24-25 D-14476 Potsdam Germany Karl Franzens Univ Graz Inst Phys Univ Platz 5 A-8010 Graz Austria High Altitude Observ HAO 3090 Ctr Green Dr Boulder CO 80301 USA Inst Astrofis Canarias IAC Via Lactea s-n San Cristobal la Laguna 38205 Tenerife Spain Univ La Laguna Dept Astrofis San Cristobal la Laguna 38205 Tenerife Spain Max Planck Inst Sonnensystemforsch Justus Von Liebig Weg 3 D-37077 Gottingen Germany Slovak Acad Sci Astron Inst Tatranska Lomnica 05960 Slovakia Appl AI Co Ltd AAICO C11-Bldg 2 Al Khaleej StAl MuntazahZone 1 Abu Dhabi U Arab Emirates Graz Univ Kanzelhohe Observ Solar & Environm Res Kanzelhohe 19 A-9521 Treffen Am Ossiacher Se Austria Skolkovo Inst Sci & Technol Bolshoy Blvd 30 Bld 1 Moscow 121205 Russia

Filaments are omnipresent features in the solar atmosphere. Their location, properties, and time evolution can provide important information about changes in solar activity and assist in the operational space weather forecast. Therefore, filaments have to be identified in full-disk images and their properties extracted from these images, but manual extraction is tedious and too time-consuming, and extraction with morphological image processing tools produces a large number of false positive detections. Automatic object detection, segmentation, and extraction in a reliable manner would allow for the processing of more data in a shorter time frame. The Chromospheric Telescope (ChroTel;Tenerife, Spain), the Global Oscillation Network Group (GONG), and the Kanzelh & ouml;he Observatory for Solar and Environmental Research (KSO;Austria) provide regular full-disk observations of the Sun in the core of the chromospheric H alpha absorption line. In this paper, we present a deep learning method that provides reliable extractions of solar filaments from H alpha filtergrams. First, we trained the object detection algorithm YOLOv5 with labeled filament data of ChroTel H alpha filtergrams. We used the trained model to obtain bounding boxes from the full GONG archive. In a second step, we applied a semi-supervised training approach where we used the bounding boxes of filaments to train the algorithm on a pixel-wise classification of solar filaments with u-net. We made use of the increased data set size, which avoids overfitting of spurious artifacts from the generated training masks. Filaments were predicted with an accuracy of 92%. With the resulting filament segmentations, physical parameters such as the area or tilt angle could be easily determined and studied. We demonstrated this in an example where we determined the rush-to-the pole for Solar Cycle 24 from the segmented GONG images. In a last step, we applied the filament detection to H alpha observations from KSO and demon

关键词： methods: observational methods: statistical techniques: image processing astronomical databases: miscellaneous catalogs Sun: chromosphere

来源：评论

学校读者我要写书评

暂无评论

Depth_ASPP_UNet plus +: a novel 2D ESPI phase unwrapping based on convolutional neural network

引用

MEASUREMENT SCIENCE AND TECHNOLOGY 2024年第6期35卷 065202-065202页

作者： Wang, Linlin Liang, Wenjie Xue, Wenke Zhang, Yinwen Wang, Chuanyun Gao, Qian Shenyang Aerosp Univ Artificial Intelligence Coll Shenyang 110136 Peoples R China

Phase unwrapping is a key step in optical metrology and physical optics to obtain accurate phase distributions. In practice, phase images obtained from electronic speckle pattern interferometry (ESPI) exhibit diverse and complex morphology, with significant shape variations and non-uniform densities among different individuals. This takes challenges for accurately extracting phase information and unwrapping the phase. With the progress of deep learning technology in optical image processing, real-time performance and accuracy have become concerned issues. In this paper, an ESPI phase unwrapping method based on convolutional neural network UNet++ is proposed. The proposed network combines the depthwise separable convolution (DSC), atrous spatial pyramid pooling (ASPP), defined as Depth_ASPP_UNet++. In this model, the use of DSC improves network computational efficiency and provides better feature representation capability. In addition, ASPP is introduced to pay more attention to the phase information of the phase image, and then obtain better phase unwrapping results. The experimental results show that our proposed method can obtain excellent results, especially with various of variable density, different noise levels, and different speckle sizes.

关键词： phase unwrapping deep learning convolutional network electronic speckle pattern interferometry optical metrology

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：