检索结果-内蒙古大学图书馆

Yolov9 models for cancer detection in biopsy images: segmentation on pan-cancer data

Yolov9 models for cancer detection in biopsy images: segment...

South Eastern European Design Automation, computer engineering, computer Networks and Social Media Conference (SEEDA-CECNSM)

作者： Panagiotis N. Smyrlis Aimilia Ntetska Thomai Karamitsou Foteini Dimaraki Katerina D. Tzimourta Pantelis Angelidis Markos G. Tsipouras Laboratory of Biomedical Technology and Digital Health School of Electrical and Computer Engineering University of Western Macedonia Kozani Greece

ISBN: (数字)9798331504489

ISBN: (纸本)9798331504496

This work examines the use of recent Yolo models on segmenting cancer biopsy images to highlight the tissue classes of interest. The experiments include recent YoloV9 and V8 architectures, trained on a large pan-cancer dataset, which contains examples for 19 distinct cancer cases. The models’ predictive capability was measured and comparatively studied, and detailed Panoptic Quality and precision metrics are discussed. The results obtained indicate the fine performance of recent Yolo models on cancer segmentation task, while all version 9 algorithms did surpass preceeding architectures. This study also examined the effect of the Yolo framework data augmentation methods for performance enhancement, by conducting experiments for two image sizes, 256 (original size) and 512 pixels.

关键词： Measurement YOLO Analytical models Image resolution Biological system modeling Biopsy computer architecture Data augmentation Data models Cancer

来源：评论

学校读者我要写书评

暂无评论

FAST CHARGING ELECTRIC BUSES IN MARIBOR: A COMPARATIVE ANALYSIS OF SIMULATION AND REAL-WORLD RESULTS 12th

FAST CHARGING ELECTRIC BUSES IN MARIBOR: A COMPARATIVE ANALY...

引用

12th International Conference on Environmental engineering, ICEE 2023

作者： Klemenčič, Mitja Španer, Marijan Lep, Marjan Moharić, Matej Department of Transportation Engineering Faculty of Civil Engineering Transportation Engineering and Architecture University of Maribor Smetanova 17 Maribor Slovenia Laboratory for Power Electronics Institute of Robotics Faculty of Electrical Engineering and Computer Science University of Maribor Koroška cesta 46 Maribor Slovenia

This paper presents a study of the energy consumption of a full electric bus charged at a fast-charging station with pantographs in the city of Maribor. The results of simulated and real tests on the PT line 6 are compared and analysed. The partly electrified PT line 6 provides insights on comparison between simulated energy consumption and re-al-life energy consumption, charging time and reliability of charging. The results reveal that the simulation considered several important factors like real life turnaround time, driving cycle, regenerative braking, charging time, peak hour charging time, heating/cooling systems, full passenger load as important and critical to identify the appropriate concept for charging. Technical and operational errors were identified in real world, due to connectivity failures or driver errors in positioning the vehicle, which were not considered in the simulation. The real-world data and procedure for selection of the most appropriate charging technology represent a good basis for full electrified Public Transport in Maribor. © 2023 The Author(s). Published by Vilnius Gediminas Technical University.

关键词： Electric buses

来源：评论

学校读者我要写书评

暂无评论

MIMO Virtual Array Design for mmWave 4D-Imaging Radar Sensors 31

MIMO Virtual Array Design for mmWave 4D-Imaging Radar Sensor...

引用

31st European Signal Processing Conference, EUSIPCO 2023

作者： Sichani, Nazila Karimian Alaee-Kerahroodi, Mohammad Raei, Ehsan Bhavani Shankar, M.R. Mehrshahi, Esfandiar Ghorashi, Seyyed Ali Department of Telecommunications Faculty of Electrical Engineering Shahid Beheshti University Tehran*** Iran University of Luxembourg Luxembourg Department of Computer Science and Digital Technologies School of Architecture Computing and Engineering University of East London LondonE16 2RD United Kingdom

ISBN: (纸本)9789464593600

The recently developed 4D-Imaging mmWave MIMO radars have significant advantages over conventional radar sensors because they offer a large virtual array in both azimuth and elevation. However, the physical placement of transmit and receive antennas to achieve the desired virtual array, while considering cost and efficiency is not intuitive. In this paper, we propose an optimization framework for optimally placing the transmit and receive array physical elements. The proposed framework obtains a sequence of optimal placements based on the coordinate descent by defining a desired virtual array and number of transmit and receive antenna elements as the objective. The objective function is then monotonically minimized in each iteration of the proposed algorithm, converging to a solution that guarantees the desired number of transmit and receive antenna elements, while the obtained virtual array is as close to the desired virtual array as possible. The simulation results are validated by simulating the virtual array of multiple commercially available 4D-Imaging radar products. © 2023 European Signal Processing Conference, EUSIPCO. All rights reserved.

关键词： Radar imaging

来源：评论

学校读者我要写书评

暂无评论

Enhancing Resolution with LSTM-RNN-based Band Fusion in Microwave Photonic Radar Systems 18

Enhancing Resolution with LSTM-RNN-based Band Fusion in Micr...

引用

Terahertz, RF, Millimeter, and Submillimeter-Wave Technology and Applications XVIII 2025

作者： Ashimbayeva, Aigerim Islam, S.M. Rezwanul Parajuli, Hum Nath Ukaegbu, Ikechi Augustine Molardi, Carlo Nakarmi, Bikash Electrical & Computer Engineering Department SEDS Nazarbayev University Astana010000 Kazakhstan Key Laboratory of Radar Imaging and Microwave Photonics Ministry of Education Nanjing University of Aeronautics and Astronautics Nanjing210016 China Integrated Device Solutions & Nanophotonics Laboratory School of Engineering and Digital Sciences Nazarbayev University Astana010000 Kazakhstan Division of Engineering Technology University of West Alabama AL35470 United States

ISBN: (纸本)9781510684782

Due to the widespread use of radar in multiple application sectors, the RF spectra allocated for radar have become congested. This leads to a shortage of adequate frequency spectrum availability, despite the capability of photonics-based systems to generate broadband radar signals. Thus, the efficient utilization of the available narrowband radar spectrum with finer resolution on object detection takes on paramount importance for radar performance in multi-radar multi-band systems, which is achieved by the fusion of multi-narrowband radar signals. In this paper, we propose and demonstrate the long short-term memory recurrent neural network (LSTM-RNN) based multi narrow-bands fusion technique. For the proof-of-concept demonstration, we generate two narrow-band and one full-band linear frequency modulated photonic radar signals respectively with the frequency range of 75 GHz - 77 GHz, 79 GHz - 81 GHz, and 75 GHz - 81 GHz utilizing heterodyning of two lasers. These signals are used to detect two objects with varying distances between the radar and objects. For training the LSTM-RNN, the fused beat signal with a gap is utilized as inputs, with the corresponding labels being a full-band beat signal. In the training stage, we optimized hyperparameters such as the number of layers, number of nodes per layer, learning rate and others using the grid search method. In the test stage, we use fused beat signals as input signals, which are distinct from the training stage. The proposed LSTM-RNN-based model showed a range resolution value of 2.6 cm for fused narrow-band signals, which is close to the theoretical range resolution of 2.5 cm. Furthermore, the peak-to-sidelobe level (PSL) improvement of 10 dB is obtained in the case of the use of LSTM-RNN and without using it for the fused narrowband signals. This ensures the effectiveness of the proposed LSTM-RNN technique for enhancing the range resolution in photonics-based multi-band radar systems. © 2025 SPIE.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Reservoir Computing Based Recurrent Neural Network for Mitigating Interference in Multi-radar Environment of Photonic Radars 18

Reservoir Computing Based Recurrent Neural Network for Mitig...

引用

Terahertz, RF, Millimeter, and Submillimeter-Wave Technology and Applications XVIII 2025

作者： Parajuli, Hum Nath Ashimbayeva, Aigerim Islam, S.M. Rezwanul Ukaegbu, Ikechi Augustine Nakarmi, Bikash Molardi, Carlo Integrated Device Solutions & Nanophotonics Laboratory School of Engineering and Digital Sciences Nazarbayev University Astana010000 Kazakhstan Electrical & Computer Engineering Department SEDS Nazarbayev University Astana010000 Kazakhstan Key Laboratory of Radar Imaging and Microwave Photonics Ministry of Education Nanjing University of Aeronautics and Astronautics Nanjing210016 China Division of Engineering Technology University of West Alabama AL35470 United States

ISBN: (纸本)9781510684782

With the growing use of radars in different applications from defense, automobiles and consumer products, external radars operating in the same or nearby frequency bands can introduce multi-interference effect and severely impact the performance of the victim radar (VR). In this paper, we present reservoir computing-based recurrent neural network (RC-RNN) to overcome the problem of interference in a multi-radar environment. Among various forms of recurrent neural networks (RNN), RC-RNN has attracted attention due to its design having lower number of parameters compared to others while still providing reasonable accuracy. For the proof-of-concept demonstration, we generated a linear frequency modulated (LFM) signal with 8 GHz bandwidth for VR using an optical injection technique in a semiconductor. We used a VR signal in the object detection experiment with two target objects with and without interference radars. To analyze and mitigate the multi-interference effect in the received echo signal, we created 100 synthetic and experimental random interference signals with varying bandwidths and chirp rates. The received signals with interferences are trained with RC-RNN. In the training stage, we optimized reservoir size, spectral radius, sparsity, and noise for regularization using the grid search method. In the test stage, 1-8 random coherent and non-coherent interference signals distinct from those used in the training stage are evaluated on the trained model. The use of RC-RNN successively achieves a mean R2 value of >92%, and a mean square error (MSE) value of © 2025 SPIE.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Machine learning-based object classification using mm-wave radar at 77 GHz - 81 GHz 18

Machine learning-based object classification using mm-wave r...

引用

Terahertz, RF, Millimeter, and Submillimeter-Wave Technology and Applications XVIII 2025

作者： Zhumabekov, Yertas Ashimbayeva, Aigerim Parajuli, Hum Nath Islam, S.M. Rezwanul Nakarmi, Bikash Molardi, Carlo Ukaegbu, Ikechi Augustine Electrical & Computer Engineering Department SEDS Nazarbayev University Astana010000 Kazakhstan Integrated Device Solutions & Nanophotonics Laboratory School of Engineering and Digital Sciences Nazarbayev University Astana010000 Kazakhstan Key Laboratory of Radar Imaging and Microwave Photonics Ministry of Education Nanjing University of Aeronautics and Astronautics Nanjing210016 China Division of Engineering Technology University of West Alabama AL35470 United States

ISBN: (纸本)9781510684782

In this paper, we experimented with k-nearest neighbors (KNN), support vector machines (SVM), and random forests (RF) machine learning (ML) algorithms to classify multiple-shape object detection using mm-wave radar. For the demonstration purpose, we generated a linear frequency modulated (LFM) radar signal at 77 GHz to 81 GHz with bandwidth ranges of 1 GHz – 4 GHz and detected flat, cylindrical, and spherical objects of various shapes and dimensions. The beat signals are recorded by shifting the radar transmitter in a step of 1 cm. The 1D range-FFT of thus obtained beat signal power levels are recorded for various radar shifts and the required data set is prepared. Acquired power level data sets are divided into 80% training and 20% test data sets through random permutation. The trained model for each ML algorithm is evaluated in the test stage. Through the analysis, the KNN and SVM provided about 70.59% accuracy and RF provided about 64.71% accuracy in recognizing different types of objects. The presented approach with a lower bandwidth radar signal, can be performed with a photonically generated high bandwidth radar signal which provides finer resolution in object detection which is under investigation. It is anticipated that with photonically generated high-bandwidth radar signal (>8 GHz), an accuracy percentage as close to 90% can be obtained. © 2025 SPIE.

关键词： Frequency modulation

来源：评论

学校读者我要写书评

暂无评论

Optical sorting:past,present and future

引用

光：科学与应用（英文版） 2025年第5期14卷 1188-1234页

作者： Meng Yang Yuzhi Shi Qinghua Song Zeyong Wei Xiong Dun Zhiming Wang Zhanshan Wang Cheng-Wei Qiu Hui Zhang Xinbin Cheng Institute of Precision Optical Engineering School of Physics Science and EngineeringTongji UniversityShanghai 200092China MOE Key Laboratory of Advanced Micro-Structured Materials Shanghai 200092China Shanghai Institute of Intelligent Science and Technology Tongji UniversityShanghai 200092China Shanghai Frontiers Science Center of Digital Optics Shanghai 200092China Tsinghua Shenzhen International Graduate School Tsinghua UniversityShenzhen 518055China Institute of Fundamental and Frontier Sciences University of Electronic Science and Technology of ChinaChengdu 610054China Department of Electrical and Computer Engineering National University of SingaporeSingapore 117583Singapore

Optical sorting combines optical tweezers with diverse techniques,including optical spectrum,artificial intelligence(AI)and immunoassay,to endow unprecedented capabilities in particle *** comparison to other methods such as microfluidics,acoustics and electrophoresis,optical sorting offers appreciable advantages in nanoscale precision,high resolution,non-invasiveness,and is becoming increasingly indispensable in fields of biophysics,chemistry,and materials *** review aims to offer a comprehensive overview of the history,development,and perspectives of various optical sorting techniques,categorised as passive and active sorting *** begin,we elucidate the fundamental physics and attributes of both conventional and exotic optical *** then explore sorting capabilities of active optical sorting,which fuses optical tweezers with a diversity of techniques,including Raman spectroscopy and machine ***,we reveal the essential roles played by deterministic light fields,configured with lens systems or metasurfaces,in the passive sorting of particles based on their varying sizes and shapes,sorting resolutions and *** conclude with our vision of the most promising and futuristic directions,including Al-facilitated ultrafast and bio-morphology-selective *** can be envisioned that optical sorting will inevitably become a revolutionary tool in scientific research and practical biomedical applications.

关键词： present future past optical sorting

来源：评论

学校读者我要写书评

暂无评论

HLSDataset: Open-Source Dataset for ML-Assisted FPGA Design using High Level Synthesis

arXiv

引用

arXiv 2023年

作者： Wei, Zhigang Arora, Aman Li, Ruihao John, Lizy The Laboratory for Computer Architecture Department of Electrical and Computer Engineering The University of Texas at Austin Austin United States

Machine Learning (ML) has been widely adopted in design exploration using high level synthesis (HLS) for faster resource, timing and power estimation at very early stages for FPGA-based design. To perform prediction accurately, high-quality and large-volume datasets are required for training ML models. However, the current datasets used in this domain are proprietary or limited in use, and practitioners have to generate their own dataset to train HLS-related ML models. This paper presents a dataset for ML-assisted FPGA design using HLS, called HLSDataset. The dataset is generated from widely used HLS C benchmarks including Polybench, Machsuite, CHStone and Rossetta. The Verilog samples are generated with a variety of directives including loop unroll, loop pipeline, and array partition to make sure optimized and realistic designs are covered. The total number of generated Verilog samples is nearly 9,000 per FPGA type. The dataset repository includes CSV (comma separated values) files containing both HLS and implementation metrics which can be easily consumed by ML model. We also include original C source code with directives, Verilog designs, post-HLS reports, post-implementation reports for each sample in the dataset, so that any metrics not present in the CSV can be easily extracted. In order to extend the dataset for future benchmarks, generation and extraction scripts are also provided. To demonstrate the effectiveness of our dataset, we undertake case studies to perform power estimation and resource usage estimation with ML models trained with our dataset. All the code and dataset are public at our github page1. We believe that HLSDataset can save valuable time for researchers by avoiding the tedious process of running tools, scripting and parsing files to generate the dataset, and enable them to spend more time where it counts, that is, in training ML models. © 2023, CC BY-NC-ND.

关键词： High level synthesis

来源：评论

学校读者我要写书评

暂无评论

LanCOPE: Language-Guided Category-Level Object Pose Estimation from a Single RGB Image

引用

IEEE Robotics and Automation Letters 2025年第7期10卷 7555-7562页

作者： Yang, Hui Sun, Wei Liu, Jian Zheng, Jin Dai, Zhenqi Mian, Ajmal Hunan University National Engineering Research Center for Robot Visual Perception and Control Technology College of Electrical and Information Engineering State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body Changsha 410082 China Central South University School of Architecture and Art Changsha 410082 China The University of Western Australia Department of Computer Science and Software Engineering 6009 WA Australia

Monocular RGB-based category-level object pose estimation is more practical and cost-effective for robotics. However, existing methods do not fully exploit the rich semantic and contextual information in multimodal data (e.g. language) that provides additional object attributes to guide the model in extracting category features more reliably. We propose a language-guided category-level object pose estimation method (LanCOPE), taking a single RGB image as input. Our method uses DINOv2 to recover depth from a single RGB image and converts it into point cloud to perceive the object's geometry. We then introduce language descriptions for the RGB image, estimated point cloud and overall scene to better guide the point cloud encoder and image encoder in learning category features. We develop a cross-modal differential perception feature fusion network to fuse multimodal features. This network employs a differential perception module to eliminate redundant information across different modalities, highlighting signifcant semantic differences and similarities. Furthermore, it uses a cross-attention mechanism to fuse the semantic information of the language and vision features, improving the overall perception. Finally, we design a denoising network based on the skip fusion transformer to recover the object pose accurately. Extensive experiments on REAL275 and Wild6D datasets show that LanCOPE achieves state-of-the-art performance. Our code is available at LanCOPE. © 2016 IEEE.

关键词： category-level object pose estimation languag RGB image

来源：评论

学校读者我要写书评

暂无评论

Blockchain-Based Secure Crowdsourcing in Wireless IoT

引用

Journal of Communications and Information Networks 2022年第1期7卷 23-36页

作者： Daquan Feng Long Zhang Shengli Zhang Qihui Wu Xianggen Xia Guangdong Province Engineering Laboratory for Digital Creative Technology Shenzhen UniversityShenzhen 518060China National Key Laboratory of Science and Technology on Communications University of Electronic Science and Technology of ChinaChengdu 611731China College of Electronic and Information Engineering Nanjing University of Aeronautics and AstronauticsNanjing 211106China Department of Electrical and Computer Engineering University of DelawareNewarkDE 19716USA

With the explosive growth of interconnected smart devices and sensors,the Internet has been entering the Internet of things(IoT)era and revolutionizing many aspects of our daily life. Meanwhile, crowdsourcing has been considered as a promising technology to realize collaborative intelligence. Therefore, more and more IoT-based crowdsourcing applications are emerged to take advantages of the widely distributed IoT devices to sense, collect, and analyze data with the aim to solve complex and nontrivial tasks. However,there exist many technical challenges to be addressed in the IoT-based crowdsourcing, such as security, privacy, and incentive provision. In this paper, we propose a blockchain-based architecture as an integrated solution to realize the secure and trustworthy crowdsourcing in wireless IoT. We first overview the challenges in the traditional crowdsourcing system. Then, we briefly introduce the background of the blockchain and smart contract, and propose a blockchain-based crowdsourcing architecture. In particular, we elaborate the utilization of smart contract on the specific phases of crowdsourcing. By deploying the smart contract instance, we confirm the proposed blockchain-based architecture is feasible.

关键词： crowdsourcing blockchain wireless IoT security privacy incentive

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：