检索结果-内蒙古大学图书馆

Assessment of patch-based mammogram denoising methods using virtual clinical trials and deep learning: trade-off between denoising strength and preservation of structural details

Assessment of patch-based mammogram denoising methods using ...

引用

Conference on Medical Imaging - Physics of Medical Imaging

作者： Dong, Vincent Maidment, Tristan D. Borges, Lucas R. Barufaldi, Bruno Ng, Susan Maidment, Andrew D. A. Real Time Tomog LLC Villanova PA 19085 USA Univ Penn Dept Radiol Philadelphia PA 19104 USA

ISBN: (数字)9781510649385

ISBN: (纸本)9781510649385;9781510649378

Digital mammography (DM) is widely used for early breast cancer detection, where high quality, low-noise images are essential for screening. To provide radiologists with the highest quality images for diagnosis, denoising techniques including deep learning (DL) have been proposed in an attempt to suppress noise while preserving structural detail. DL model training depends on access to large training datasets with ground truth data to capture features relevant to the task. Ground truth, in the form of x-ray quantum-noise free DM images, is not available, so the use of computer simulations provides a viable method for this to be addressed. This study proposes a patch-based DL convolutional neural network (CNN) model for mammography denoising, trained on data obtained through our lab's virtual clinical trial (VCT) framework (OpenVCT). We extracted patches from VCT samples and trained a deep residual CNN to distinguish the patterns of noise from the underlying structure of the DM. Different denoising model configurations and combinations of denoising loss functions were explored to learn the simulated noise mappings and fine-tune the denoising performance to identify the trade-offs between denoising strength and fine structure preservation. We evaluated the denoising performance of these models by comparing the change in several image quality metrics, including peak signal-to-noise ratio (PSNR) and power spectra (PS). Our results indicate that our trained DL denoising model can detect and remove noise while preserving the original structure, resulting in a denoised reconstruction of simulated mammograms.

关键词： Digital mammography deep learning denoising residual convolutional neural network virtual clinical trial

来源：评论

学校读者我要写书评

暂无评论

Research on YOLOv8-Based Household Item Detection for Indoor Scenes

Research on YOLOv8-Based Household Item Detection for Indoor...

引用

Data Intelligence and Innovative Application (DIIA), IEEE International Conference on

作者： Jingwen Wu Xiaonan Luo Guangxi Key Laboratory of Image and Graphic Intelligent Processing Guilin University of Electronic Technology Guilin China

ISBN: (数字)9798331504434

ISBN: (纸本)9798331504441

This research is driven by the growing need for efficient, scalable object detection in smart home applications with real-time performance and resource constraints. In this study, we utilize the YOLOv8 deep learning model for household item detection, focusing on its performance in real-time scenarios. We compare different versions of YOLOv8—specifically, YOLOv8n, YOLOv8s, and YOLOv8m—using a preprocessed custom dataset. The results indicate that YOLOv8n stands out due to its lower parameter count and faster inference speed, making it ideal for real-time applications, even though it has slightly lower precision and recall. In contrast, while YOLOv8s and YOLOv8m show improvements in certain metrics, their higher computational demands limit their effectiveness in resource-constrained settings. Overall, YOLOv8n effectively balances speed and accuracy, offering a viable solution for household item detection.

关键词： Training Measurement Analytical models Accuracy Computational modeling Heuristic algorithms Smart homes Object detection real-time systems Tuning

来源：评论

学校读者我要写书评

暂无评论

real-time crack assessment using deep neural networks with wall-climbing unmanned aerial system

引用

COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING 2020年第6期35卷 549-564页

作者： Jiang, Shang Zhang, Jian Southeast Univ Jiangsu Key Lab Engn Mech Nanjing Jiangsu Peoples R China Southeast Univ Sch Civil Engn Nanjing Jiangsu Peoples R China

Crack information provides important evidence of structural degradation and safety in civil structures. Existing inspection methods are inefficient and difficult to rapidly deploy. A real-time crack inspection method is proposed in this study to address this difficulty. Within this method, a wall-climbing unmanned aerial system (UAS) is developed to acquire detailed crack images without distortion, then a wireless data transmission method is applied to fulfill real-time detection requirements, allowing smartphones to receive real-time video taken from the UAS. Next, an image data set including 1,330 crack images taken by the wall-climbing UAS is established and used for training a deep-learning model. For increasing detection speed, state-of-the-art convolutional neural networks (CNNs) are compared and employed to train the crack detector;the selected model is transplanted into an android application so that the detection of cracks can be undertaken on a smartphone in real time. Following this, images with cracks are separated and crack width is calculated using an image processing method. The proposed method is then applied to a building where crack information is acquired and calculated accurately with high efficiency, thus verifying the practicability of the proposed method and system.

关键词： deep neural networks

来源：评论

学校读者我要写书评

暂无评论

A Tiny Efficient U-Net with Gated Linear Attention for Medical image Segmentation

A Tiny Efficient U-Net with Gated Linear Attention for Medic...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Sibo Ju Zhaozhen Chen Xiangwen Liao Yiqing Shen Junjun He Yanzhou Su College of Computer and Data Science Fuzhou University Fuzhou China Department of Computer Science Johns Hopkins University Baltimore USA Shanghai Artificial Intelligence Laboratory Shanghai China

ISBN: (数字)9798350386226

ISBN: (纸本)9798350386233

Medical image segmentation is crucial for diagnosis and treatment planning. While recent advancements in deep learning, particularly UNet variants, have improved segmentation performance, they often result in increased model complexity, which limits their real-time applicability on resource-constrained devices in clinical settings. To address this challenge, we present the Tiny Efficient U-Net (TE-UNet), a novel lightweight model balancing efficiency and accuracy. TE-UNet uses a U-shaped encoder-decoder framework with a gated linear attention mechanism to process low-level and high-level features, preserving details and reducing complexity. It employs depth-wise separable convolutions for higher-level processing, enhancing efficiency without losing performance. Additionally, skip connections improve multi-scale feature extraction and information flow. Experiments on two public datasets across different modalities demonstrate that TE-UNet outperforms ten state-of-the-art methods, maintaining a parameter size under 40KB and low computational cost. TE-UNet makes real-time segmentation more accessible for various clinical applications.

关键词： Performance evaluation image segmentation Attention mechanisms Accuracy Logic gates real-time systems Computational efficiency Complexity theory Planning Medical diagnostic imaging

来源：评论

学校读者我要写书评

暂无评论

Towards the spatial analysis of motorway safety in the connected environment by using explainable deep learning

引用

KNOWLEDGE-BASED SYSTEMS 2023年 269卷

作者： Greguric, Martin Vrbanic, Filip Ivanjko, Edouard Univ Zagreb Fac Transport & Traff Sci Vukeliceva St 4 HR-10000 Zagreb Croatia

This study investigates the application of Connected and Automated Vehicles (CAVs) as moving sensors that transmit their speed and position in real-time for spatial analysis of motorway safety. Those data are used for the generation of image-alike inputs which describe the speed distribution over the entire motorway model in the form of heat-maps. Their labels are safety categories computed by using average time-to-Collision (TTC). The Convolution Neural Network (CNN) is proposed to predict the category of safety based on the image-alike labeled dataset. Furthermore, Explainable Artificial Intelligence (xAI) is used to explain which segments of image-alike inputs are critical for the accurate prediction of safety. It is applied to selected inputs with the best learning performance and if they represent the undesirable safety categories. The study investigates the impact of various penetration rates of CAVs with the Intelligent Speed Adaptation (ISA) system on the spatial distribution of safety- critical regions. The higher penetration rates of the CAVs with the ISA system reduce the dispersion and intensity of critical regions computed by xAI over the entire motorway. Those regions are located at the most critical part of the analyzed motorway segment where the on-ramps flow interacts with the mainstream flow and its adjacent off-ramp. The higher penetration rate of CAVs with the ISA system induces a more consistent and localized distribution of critical regions regarding safety. Thus, this confirms that critical regions for safety categorization computed by xAI correspond with the motorway region with the most critical safety.(c) 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://***/licenses/by-nc-nd/4.0/).

关键词： Connected and Automated Vehicles Explainable Artificial Intelligence deep learning Traffic safety analysis Variable Speed Limit Intelligent Speed Adaptation

来源：评论

学校读者我要写书评

暂无评论

Desert mixed seismic noise suppression by using multiple forward models and a supervised deep-learning method

引用

EXPLORATION GEOPHYSICS 2021年第4期52卷 431-445页

作者： Dong, Xintong Wang, Hongzhou Zhong, Tie Li, Yue Jilin Univ Dept Informat Engn Changchun Peoples R China Northeast Elect Power Univ Dept Commun Engn Jilin Jilin Peoples R China

Desert seismic data pose a particular problem due to its low signal-to-noise ratio and severe spectral overlapping. Most of the existing denoising methods have negative suppression effect on the desert mixed seismic noise, which mainly includes random noise and surface waves. The convolutional neural network (CNN) is a classical supervised deep-learning method that has proven to be an effective tool for noise suppression. The drawback is that CNN usually needs a large amount of ideal noise-free data to complete network training, which largely limits its applications to seismic data denoising. In this paper, we combine forward modelling with CNN to overcome this technical problem. Specifically, we construct multiple forward models from different physical parameters to obtain a noise-free seismic dataset by finite difference method. After adding noise data taken from a real desert noise record, we use the above data as the input for CNN training. Then, the MSE loss function allows us to obtain an optimal denoising model for the mixed noise suppression of desert seismic data. Continuing with this optimal model, we can predict mixed noise, which, once subtracted from the desert original seismic data, provide the denoised desert seismic data. Experiments with synthetic and real data demonstrate that this CNN-based denoising method can suppress the desert mixed seismic noise successfully. Several tests using artificially generated models with a different number of signal patches prove that the forward modelling can significantly improve the continuity of events and prevent the appearance of false events. In addition, analysis of the effect of network depth supports the effectiveness of our method through a trade-off among higher signal-to-noise ratio, lower root mean square error and shorter training time.

关键词： image processing low frequency seismic exploration reflection

来源：评论

学校读者我要写书评

暂无评论

Automated Detection and Detterents of Harmful Birds in Tomato Cultivation

Automated Detection and Detterents of Harmful Birds in Tomat...

引用

8th International Conference on Inventive Computation Technologies, ICICT 2025

作者： Anuja Beula, J. Gowrisanker, M. Department of Computer Science and Engineering KIT-Kalaignarkarunanidhi Institute of Technology Coimbatore India Department of Computer Science and Engineering KIT-KalaignarKarunanidhi Institute of Technology Coimbatore India

ISBN: (数字)9798331512248

ISBN: (纸本)9798331512255

This work introduces an autonomous learning system for preventing bird damage in tomato crops based on state-of-the-art computer vision and machine learning. The system applies Convolutional Neural Networks (CNN) and Random Forest (RF) for real-time bird detection and classification. Ensemble learning enhances accuracy, minimizing crop loss at the cost of being cost-effective. The research compares different methods, such as deep learning-based deterrent systems and Internet of Things (IoT) solutions. Comparison with conventional bird-scaring methods is made to show the benefits of automation. Experimental evidence confirms the effectiveness of the system and proves its potential application in precision agriculture. The improvements for the future involve the use of Unmanned Aerial Vehicles (UAVs) and IoT-based interventions to further improve scalability. © 2025 IEEE.

关键词： Automated Disease Detection Avian Identification Ensemble Classification Model image processing Precision Agriculture

来源：评论

学校读者我要写书评

暂无评论

EDGE detection and object identification for vision enhancement at rescue operations using deep learning techniques 12

EDGE detection and object identification for vision enhancem...

引用

12th International Conference on Computer Communication and Informatics, ICCCI 2022

作者： Selvakarthi, D. Sivabalaselvamani, D. Ashwath, S. Kalaivanan, A.P. Aswin Manikandan, K. Pradeep, C. Kongu Engineering College Department of Electronics and Instrumentation Engineering Tamilnadu Erode638060 India Kongu Engineering College Department of Computer Applications Erode Tamilnadu Perundurai638060 India Kongu Engineering College Department of Eie Erode Tamilnadu Perundurai638060 India

ISBN: (纸本)9781665480352

In this modern world the rate of occurrence of accidents and military operation in dark and unpleasant goes on increasing. At this situation the rescue squad faces difficulty in navigation. The visual detection and object clarity is difficult in risky situations like rescue operations in dark environment, secret armed rescue and military based operations. Due to this difficulty the failure rate of the operation is high. The proposed idea based on AI -deep learning technology provides better clarification on edge detection and enhances the vision with the help of OpenCV (Computer Vision Library) and other frameworks. Data transmission and image processing is implemented with the help of raspberry pi. The object identification is implemented in the dark environment. This makes the navigation of the rescue quad inside the operation make easier. To provide visual enhancement for the structures of buildings and objects during rescue operation and also the visual projection of object structures in secret operations at military. The real time object edges and classification of objects and internal structures of environment is projected based on the user comfortless. © 2022 IEEE.

关键词： Failure analysis

来源：评论

学校读者我要写书评

暂无评论

Polarimetric HRRP recognition based on feature-guided Transformer model

引用

ELECTRONICS LETTERS 2021年第18期57卷 705-707页

作者： Zhang, Liang Han, Chang Wang, Yanhua Li, Yang Long, Teng Beijing Inst Technol Sch Informat & Elect Radar Res Lab Beijing 100081 Peoples R China Beijing Inst Technol Chongqing Innovat Ctr Chongqing 401147 Peoples R China Beijing Key Lab Embedded Real Time Informat Proc Beijing 100081 Peoples R China

Polarimetric high-resolution range profile (HRRP) holds great potential for radar automatic target recognition (RATR) owing to its capability of providing both polarimetric and spatial scattering information. In recent years, deep learning (DL) has obtained state-of-the-art results in many classification tasks and has drawn great attention in the RATR field. However, as one of the most challenging tasks in RATR, small training sample case will restrict the application of DL because its superior performance generally depends on a large number of training samples. A feature-guided deep model based on Transformer framework is proposed for polarimetric HRRP recognition with limited training samples. In the proposed model, artificial features are introduced to the attention module to guide the model focus on the range cells of HRRP with more target scattering information so as to reduce the dependence of the model on the number of training samples. Several different approaches are also studied to measure the similarity between artificial features and HRRP data to further improve the learning capacity of the model. Experimental results demonstrate that the proposed feature-guided Transformer model modifying by Cosine similarity measure is able to achieve a better performance for polarimetric HRRP recognition with limited training samples.

关键词： image recognition Radar equipment, systems and applications Computer vision and image processing techniques Electrical engineering computing Machine learning (artificial intelligence)

来源：评论

学校读者我要写书评

暂无评论

The Workpiece Sorting Method Based on Improved YOLOv5 For Vision Robotic Arm 19

The Workpiece Sorting Method Based on Improved YOLOv5 For Vi...

引用

19th IEEE International Conference on Mechatronics and Automation (IEEE ICMA)

作者： Ma, Zhongli Zeng, Yuehan Zhang, Linshuai Li, Jiadi Chengdu Univ Informat Technol Dept Control Engn CULT Chengdu Sichuan Peoples R China

ISBN: (数字)9781665408530

ISBN: (纸本)9781665408530;9781665408523

In order to solve the problem of high error rate and poor real-time performance in the workpieces sorting process for traditional industrial robotic arms, this paper designed a vision robotic arm testing platform with real-time processing ability, and proposes a kind of workpieces sorting method based on improved YOLOv5 used to the vision robotic arm. By replacing the focus layer in the YOLOv5 backbone network, embedding the coordinate attention module, which re-weights the feature maps from the channel and spatial, improves the object detection accuracy of the YOLOv5 model. The workpiece sorting test platform consists of an NVIDIA Jetson nano controller and a vision robotic arm. The hand-eye calibration of the robotic arm is completed by the Zhang Zhengyou calibration method and the TsarLenz method. The workpiece target image was collected, tagged and data augmented to create the target workpiece dataset. And use TensorRT to optimize the inference acceleration of the model to adapt to the hardware platform requirements. The test shows that the improved YOLOv5 model can well ensure the stable operation of the test platform, and improve the accuracy and real -time performance of workpiece target recognition.

关键词： deep learning Robotic arm sorting YOLO v5 Object detection Attention mechanism TensorRT

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：