检索结果-内蒙古大学图书馆

IS and T International Symposium on Electronic Imaging 2024: 22nd image processing: Algorithms and Systems, IPAS 2024

作者： Sequeira, Aaron Sam, Febin Jain, Anshu Swami, Pramod National Institute of Technology Karnataka India National Institute of Technology Dimapur India Indian Institute of technology Bombay India Texas Instruments United States

deep learning has enabled rapid advancements in the field of image processing. learning based approaches have achieved stunning success over their traditional signal processing-based counterparts for a variety of applications such as object detection, semantic segmentation etc. This has resulted in the parallel development of hardware architectures capable of optimizing the inferencing of deep learning algorithms in real time. Embedded devices tend to have hard constraints on internal memory space and must rely on larger (but relatively very slow) DDR memory to store vast data generated while processing the deep learning algorithms. Thus, associated systems have to be evolved to make use of the optimized hardware balancing compute times with data operations. We propose such a generalized framework that can, given a set of compute elements and memory arrangement, devise an efficient method for processing of multidimensional data to optimize inference time of deep learning algorithms for vision applications. © 2024, Society for Imaging Science and Technology.

关键词： learning algorithms

来源：评论

学校读者我要写书评

暂无评论

HFD-NET: a real-time deep learning-based visual algorithm with weak feature enhancement for hair follicle detection

引用

SIGNAL image AND VIDEO processing 2025年第5期19卷 1-15页

作者： Zhang, Liyang Tong, Juntao Chen, Yue Wang, Nian Li, Weiang Chen, Mengxin Tian, Jinghong Key Lab Urban Rail Transit Intelligent Operat & Ma Jinhua 321000 Peoples R China Zhejiang Normal Univ Coll Comp Sci & Technol Jinhua 321000 Peoples R China Zhejiang Normal Univ Coll Engn Jinhua 321000 Peoples R China

Hair follicle detection in complex scalp environments remains challenging due to small target sizes, morphological similarities, and background interference. To address these issues, we propose HFD-NET, a novel real-time detection framework that significantly enhances weak feature representation while maintaining computational efficiency. Unlike existing methods, HFD-NET introduces three key innovations: (1) the CSRFConv module, which dynamically fuses spatial-channel features to suppress background noise and amplify discriminative follicle characteristics;(2) the C3k2_IBC module, optimized for multi-scale small-object detection through inverted bottleneck convolutions;and (3) the ADown module, which minimizes information loss during downsampling by preserving critical edge and texture details. Extensive experiments on the FDU_HairFollicleDataset demonstrate HFD-NET's superiority, achieving a 67.1% mAP@0.5-outperforming YOLO11n by 5.5%-with only a 0.1M parameter increase and negligible computational overhead. HFD-NET also generalizes effectively to public datasets (e.g., 71.6% mAP@0.5 on HFDC), surpassing Faster R-CNN, SSD, and recent YOLO variants in accuracy-efficiency trade-offs. This work bridges the gap between high-precision detection and real-time applicability, offering a practical solution for clinical hair transplantation and scalp health monitoring.

关键词： Hair follicle detection YOLO11 deep learning Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Robust Point Tracking Based On image Matching And Machine learning On Video images Taken from Fast Moving Camera 32

Robust Point Tracking Based On Image Matching And Machine Le...

引用

32nd IEEE Signal processing and Communications Applications Conference (SIU)

作者： Guven, Ali Yetik, Imam Samil TOBB Ekon & Teknol Univ Elekt & Elekt Muhendisligi Bolumu Ankara Turkiye

ISBN: (纸本)9798350388978;9798350388961

Recently, providing real-time navigation of unmanned aerial vehicles independent of global positioning systems has become of great importance. The state-of-the-art methods based on deep learning, which give good results in certain datasets, and the existing methods can not provide real-time and good solutions on images with dynamic and fast moving. Moreover, the methods, were developed so far, were focused on object-based tracking algorithms. In this paper, the tracking of the points belonging to the target pattern, found by image matching, was performed with the machine learning model we developed for 10 sequential video images. The features extracted for the machine learning model are: (i) the change between the points of the previous image and the image before that, (ii) the points of interest in the previous image, (iii) the changes found with the homography matrix between sequential images. It was experimentally shown that, point tracking can be achieved with the least error, on avarage about 23 pixels for a 2 mega-pixel resolution image, among the algorithms in the literature that can process more than 30 images per second in a CPU environment of 2 GHz or above.

关键词： image matching image homography image-object tracking image-point tracking apattern detection machine learning image processing feature extraction

来源：评论

学校读者我要写书评

暂无评论

Diagnosing the spores of tomato fungal diseases using microscopic image processing and machine learning

引用

MULtimeDIA TOOLS AND APPLICATIONS 2024年第26期83卷 67283-67301页

作者： Javidan, Seyed Mohamad Banakar, Ahmad Vakilian, Keyvan Asefpour Ampatzidis, Yiannis Rahnama, Kamran Tarbiat Modares Univ Dept Biosyst Engn Tehran Iran Gorgan Univ Agr Sci & Nat Resources Dept Biosyst Engn Gorgan Iran Univ Florida Southwest Florida Res & Educ Ctr Agr & Biol Engn Dept 2685 FL-29 Immokalee FL 34142 USA Gorgan Univ Agr Sci & Nat Resources Fac Plant Prod Dept Plant Protect Gorgan Iran

Accurate diagnosis of plant diseases by the assessment of pathogen presence to reduce disease-related production loss is one of the most fundamental issues for farmers and specialists. This will improve product quality, increase productivity, reduce the use of fungicides, and reduce the final cost of agricultural production. Today, new technologies such as image processing, artificial intelligence, and deep learning have provided reliable solutions in various fields of precision agriculture and smart farm management. In this research, microscopic image processing and machine learning have been used to identify the spores of four common tomato fungal diseases. A dataset including 100 microscopic images of spores for each disease was developed, followed by the extraction of the texture, color, and shape features from the images. The classification results using random forest revealed an accuracy higher than 98%. Besides, as a reliable feature selection algorithm, the butterfly optimization algorithm (BOA) was used to detect the effective image features to identify and classify diseases. This algorithm recognized image textural features as the most effective features in the diagnosis and classification of disease spores. Considering only the eight most effective features selected with BOA resulted in an accuracy of 95% in disease detection. To further investigate the performance of the proposed method, its accuracy was compared with the accuracies of convolutional neural networks and EfficientNet as two reliable deep learning algorithms. Not only the prediction accuracy of these methods was not favorable (65 and 83.55%, respectively), they were very time-consuming. According to the findings, the proposed framework has high reliability in disease diagnosis and can help in the management of tomato fungal diseases.

关键词： Artificial intelligence Butterfly optimization algorithm Disease diagnosis Microscopic image processing Morphological features Tomato disease spores

来源：评论

学校读者我要写书评

暂无评论

MMG-Based Motion Segmentation and Recognition of Upper Limb Rehabilitation Using the YOLOv5s-SE

引用

SENSORS 2025年第7期25卷 2257-2257页

作者： Cao, Gangsheng Jia, Shen Wu, Qing Xia, Chunming East China Univ Sci & Technol Sch Mech & Power Engn Shanghai 200237 Peoples R China Shanghai Univ Engn Sci Sch Mech & Automot Engn Shanghai 201620 Peoples R China

Mechanomyography (MMG) is a non-invasive technique for assessing muscle activity by measuring mechanical signals, offering high sensitivity and real-time monitoring capabilities, and it has many applications in rehabilitation training. Traditional MMG-based motion recognition relies on feature extraction and classifier training, which require segmenting continuous actions, leading to challenges in real-time performance and segmentation accuracy. Therefore, this paper proposes an innovative method for the real-time segmentation and classification of upper limb rehabilitation actions based on the You Only Look Once (YOLO) algorithm, integrating the Squeeze-and-Excitation (SE) attention mechanism to enhance the model's performance. In this paper, the collected MMG signals were transformed into one-dimensional time-series images. After image processing, the training set and test set were divided for the training and testing of the YOLOv5s-SE model. The results demonstrated that the proposed model effectively segmented isolated and continuous MMG motions while simultaneously performing real-time motion category prediction and outputting results. In segmentation tasks, the base YOLOv5s model achieved 97.9% precision and 98.0% recall, while the improved YOLOv5s-SE model increased precision to 98.7% (+0.8%) and recall to 98.3% (+0.3%). Additionally, the model demonstrated exceptional accuracy in predicting motion categories, achieving an accuracy of 98.9%. This method realizes the automatic segmentation of time-domain motions, avoids the limitations of manual parameter adjustment in traditional methods, and simultaneously enhances the real-time performance of MMG motion recognition through image processing, providing an effective solution for motion analysis in wearable devices.

关键词： mechanomyography upper limb motion segmentation pattern recognition target detection YOLOv5s-SE deep learning

来源：评论

学校读者我要写书评

暂无评论

Detection of Moiré pattern in high-resolution images

引用

SIGNAL image AND VIDEO processing 2024年第1期18卷 561-568页

作者： Zhang, Yu Shen, Li-Yong Univ Chinese Acad Sci Sch Math Sci Beijing 100049 Peoples R China

Despite the improvements in the quality of digital images brought about by the development of digital cameras and smartphones, taking clean photographs remains a challenge due to the presence of noise such as moire patterns, especially for high-resolution images commonly used in today's real life. To more efficiently and accurately identify and remove moires in high-resolution images, this paper proposes an improved network based on the high-resolution network HRDN. We conducted experiments on the FHDMi dataset to evaluate the effectiveness of our method. The results demonstrate that the improved algorithm significantly enhances the performance of moire detection in high-resolution images, making it better suited to meeting the demands of real-world applications.

关键词： image process Demoireing High-resolution deep learning

来源：评论

学校读者我要写书评

暂无评论

Auto-BCS: A Hybrid System for real-time Breast Cancer Screening from Pathological images

引用

JOURNAL OF IMAGING INFORMATICS IN MEDICINE 2024年第4期37卷 1752-1766页

作者： Ekta, Vandana Bhatia, Vandana Netaji Subhas Univ Technol Delhi India

Breast cancer is recognized as a prominent cause of cancer-related mortality among women globally, emphasizing the critical need for early diagnosis resulting improvement in survival rates. Current breast cancer diagnostic procedures depend on manual assessments of pathological images by medical professionals. However, in remote or underserved regions, the scarcity of expert healthcare resources often compromised the diagnostic accuracy. Machine learning holds great promise for early detection, yet existing breast cancer screening algorithms are frequently characterized by significant computational demands, rendering them unsuitable for deployment on low-processing-power mobile devices. In this paper, a real-time automated system "Auto-BCS" is introduced that significantly enhances the efficiency of early breast cancer screening. The system is structured into three distinct phases. In the initial phase, images undergo a pre-processing stage aimed at noise reduction. Subsequently, feature extraction is carried out using a lightweight and optimized deep learning model followed by extreme gradient boosting classifier, strategically employed to optimize the overall performance and prevent overfitting in the deep learning model. The system's performance is gauged through essential metrics, including accuracy, precision, recall, F1 score, and inference time. Comparative evaluations against state-of-the-art algorithms affirm that Auto-BCS outperforms existing models, excelling in both efficiency and processing speed. Computational efficiency is prioritized by Auto-BCS, making it particularly adaptable to low-processing-power mobile devices. Comparative assessments confirm the superior performance of Auto-BCS, signifying its potential to advance breast cancer screening technology.

关键词： Breast cancer screening deep learning Classification models Hybrid system On-device image recognition

来源：评论

学校读者我要写书评

暂无评论

A comprehensive study on enhanced QR extraction techniques with deep learning-based verification

引用

APPLIED INTELLIGENCE 2025年第7期55卷 1-22页

作者： Alam, Nur Sagar, A. S. M. Sharifuzzaman Zhang, Wenqi Jin, Taicheng Dosset, Arailym Dang, L. Minh Moon, Hyeonjoon Sejong Univ Dept Comp Sci & Engn Seoul 05006 South Korea Sejong Univ Dept Intelligent Mechatron Engn Seoul 05006 South Korea Duy Tan Univ Inst Res & Dev Da Nang 550000 Vietnam Duy Tan Univ Fac Informat Technol Da Nang 550000 Vietnam

In the digital age, Quick Response (QR) codes have become essential in sectors such as digital payments and ticketing, propelled by advancements in Internet of Things (IoT) and deep learning. Despite their growing use, there are significant challenges in the accurate extraction and verification of QR codes, particularly in dynamic environments. Traditional methods struggle with issues like variable lighting, complex backgrounds, and counterfeits, which degrade the performance of QR code extraction and verification processes. This paper introduces a comprehensive approach that refines QR code extraction using enhanced adaptive thresholding techniques and incorporates a deep learning framework specifically tailored for robust QR code verification. Our methodology integrates dynamic window size adjustment, statistical weighting, and post-thresholding refinement to ensure precise QR code extraction under varying conditions. The verification process employs the ShuffleNetV2 network to ensure high performance with significantly low processing times suitable for real-time applications. Additionally, our deep learning model is trained on a comprehensive dataset comprising 28,523 images across 24 distinct QR code pattern classes, including variations in lighting, noise, and backgrounds to simulate real-world conditions. Experimental results demonstrate that our proposed methodology outperforms competing techniques in both processing speed and recognition accuracy, achieving a processing time of 0.08 seconds and a validation accuracy of 99.99% in constrained scenarios. Our approach shows an exceptional ability to distinguish real QR codes from counterfeits and highlights the significance and efficacy of our method in addressing contemporary challenges.

关键词： QR code extraction deep learning Feature extraction Verification

来源：评论

学校读者我要写书评

暂无评论

Efficient Remote Health Monitoring Using deep learning and Parallel Systems

引用

INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS 2024年第10期15卷 174-186页

作者： El Khadiri, Zakaria Latif, Rachid Saddik, Amine Ibn Zohr Univ Agadir Natl Sch Appl Sci Lab Syst Engn & Informat Technol LISTI Agadir Morocco Ibn Zohr Univ Fac Appl Sci Ait Melloul Morocco

This study presents a novel approach for non- contact extraction of physiological parameters, such as heart rate and respiratory rate, from facial images captured using RGB cameras, leveraging recent advancements in deep learning and signal processing techniques. The proposed system integrates Artifacts intelligent-driven algorithms for accurately estimating vital signs, addressing key challenges such as variations in lighting conditions, facial orientation, and noise. The system is implemented on both a naive homogeneous architecture and an optimized heterogeneous CPU-GPU system to enhance real-time performance and computational efficiency. A comparative analysis is performed to evaluate processing speed, accuracy, and resource utilization across both architectures. Results demonstrate that the optimized heterogeneous system significantly outperforms the homogeneous counterpart, achieving faster processing times while maintaining high accuracy levels. This performance boost is achieved through parallel computing frameworks such as OpenMP and OpenCL, which allow for efficient resource allocation and scalability. The research highlights the potential of heterogeneous architectures for real-time healthcare applications, including remote patient monitoring and telemedicine, providing a robust solution for future developments in non-invasive health technology.

关键词： real-time healthcare embedded systems heterogeneous computing deep learning CPU-GPU architecture

来源：评论

学校读者我要写书评

暂无评论

LDFNN: A lightweight dual-feedback neural network for space target component detection

引用

ADVANCES IN SPACE RESEARCH 2025年第7期75卷 5702-5717页

作者： Zhou, Wenbo Wang, Zebin Li, Ligang Liu, Bo Wu, Yijia Chinese Acad Sci Natl Space Sci Ctr Key Lab Elect & Informat Technol Space Syst Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Comp Sci & Technol Beijing 100049 Peoples R China Shanghai Acad Spaceflight Technol Shanghai 201109 Peoples R China Beijing Univ Chem Technol Beijing 100029 Peoples R China Shandong Univ Finance & Econ Sch Comp Sci & Technol Jinan Peoples R China

As the number of spacecraft and the complexity of their missions increase, space target component detection technology is becoming critical for ensuring spacecraft safety in orbit. An effective detection system not only monitors and identifies potential collision threats in real-time, enabling timely obstacle avoidance or path adjustments, but also provides crucial localization and identification capabilities for autonomous spacecraft maintenance and repair. However, most existing methods overlook image quality degradation resulting from prolonged on-orbit operation. Although deep learning models have improved detection accuracy, they are often burdened by excessive network parameters and high computational complexity, which limits their applicability to low-power, lightweight, and resource- constrained on-board processing. In this study, we propose a lightweight dual-feedback neural network (LDFNN) for space target component detection and recognition. The network is designed to address the challenges posed by the space environment through a dual- feedback mechanism. The outer-feedback mechanism enhances recognition stability and accuracy by incorporating modulation transfer function (MTF) analysis and image quality assessment, while the inner-feedback mechanism is optimized for low-power on-orbit processing, improving both real-time performance and generalization. (c) 2025 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

关键词： Space target Component detection Dual-feedback neural network deep learning MTF

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：