检索结果-内蒙古大学图书馆

Real-time Vehicle Detection and Tracking on Fisheye Traffic Monitoring Video in Compressed Domain

APSIPA Transactions on Signal and Information Processing 2023年第4期12卷

作者： Ardianto, Sandy Hang, Hsueh-Ming Cheng, Wen-Huang Electrical Engineering and Computer Science International Graduate Program National Yang Ming Chiao Tung University Hsinchu Taiwan Institute of Electronics National Yang Ming Chiao Tung University Hsinchu Taiwan Department of Computer Science and Information Engineering National Taiwan University Taipei Taiwan

Our goal is to develop real-time vehicle detection and tracking schemes for fisheye traffic monitoring video using the temporal information in the compressed domain without decoding the entire video. Two algorithms are proposed. The first algorithm starts with a conventional single-frame detector, but we introduce a multi-frame information fusion stage to improve the final detection and tracking accuracy, which is implemented using multi-modal bi-directional LSTM (MM bi-LSTM) network. The second algorithm first constructs multi-frame motion trail image, and then a single-image multi-head detector is designed to produce bounding boxes of an individual frame. The first scheme can be viewed as a detect-to-track design, and the second scheme is track-to-detect. We tested our proposals on the ICIP2020 VIP Cup dataset in H.265 video format. The aforementioned algorithms are applied to the motion fields and residual images in the H.265 compressed data set. It turns out that their detection and tracking performances are on par with their pixel-domain counterparts, and they can achieve the state-of-the-art accuracy of conventional video object detectors and trackers. If the decoding process for video compression is not counted, their computational complexities are much lower than the conventional pixel-domain video object detectors and trackers. © 2023 S. Ardianto, H.-M. Hang and W.-H. Cheng.

关键词： Image compression

来源：评论

学校读者我要写书评

暂无评论

Reducing Errors and Powers in Lpddr for Dnn Inference: A Compression and Iecc-Based Approach

SSRN

引用

SSRN 2025年

作者： Hong, Jae-Youn Jang, Je-Woo Cho, Sung-Hyuk Kong, Youngbae Kim, Sungkyu Kang, Youngjung Ko, Jaehyung Chung, Jaeyong Yang, Joon-Sung Department of Electrical and Electronic Engineering Yonsei University Seoul Korea Republic of Memory Division Samsung Electronics Hwaseong Korea Republic of Department of System Semiconductor Engineering Yonsei University Seoul Korea Republic of BK21 Graduate Program in Intelligent Semiconductor Technology Yonsei University Seoul Korea Republic of

In modern edge systems, the demand for data processing for complex DNN tasks, is rapidly increasing. To address this, various compression schemes have been proposed to enable on-device AI while meeting the strictpower and storage constraints. However, despite these advancements, the compatibility of the compression methods with edge device memory(LPDDR), has not been thoroughly investigated. LPDDR operates at low voltage and faces reliability challenges, which is particularly concerning for applications where accuracy is critical, such as ADAS or medical devices. To address these concerns, ECC engine, known as IECC, is employed within each LPDDR bank. While IECC improves reliability, it also incurs performance penalties due to Read-Modify-Write(RMW) operations and parity storage overheads. This paper introduces RELIA, a DNN weight compression scheme with three-stage protection, aimed at enabling power-efficient and reliable DNN operations in mobile environments. RELIA reduces the operation granularity of the IECC engine from SEC(136,128) to SEC(72,64), eliminating RMW overhead. Additionally, it proposes SEC-FOEC(72,64) scheme that can correct 99.97% of LPDDR errors. To mitigate the added storage overhead, compression scheme based on DNN weight characteristics is introduced. Experimental results show RELIA reduces power, cycle, energy, and storage by 17.35%, 13.2%, 32.7%, and 23.6%, respectively, while offering superior reliability. © 2025, The Authors. All rights reserved.

关键词： Biomedical equipment

来源：评论

学校读者我要写书评

暂无评论

Energy-aware Model Predictive Control with Obstacle Avoidance

Energy-aware Model Predictive Control with Obstacle Avoidanc...

引用

2021 International Conference on Unmanned Aircraft Systems, ICUAS 2021

作者： Santos, Marcelo A. Ferramosca, Antonio Raffo, Guilherme V. Federal University of Minas Gerais Graduate Program in Electrical Engineering Belo Horizonte Brazil University of Bergamo Department of Management Information and Production Engineering Dalmine24044 Italy Federal University of Minas Gerais Department of Electronics Engineering Belo Horizonte Brazil

ISBN: (纸本)9780738131153

This work proposes a single-layer finite-horizon optimal control strategy to solve the autonomous navigation problem while accounting for energy efficiency and providing obstacle avoidance feature in cluttered environments with unknown obstacles. Considering the rate capacity effect of electric batteries, the nonlinear state-of-charge behavior is described and included in the optimal control problem to achieve energy-awareness. Besides, artificial potential fields are considered to obtain obstacle avoidance capabilities. The control problem is formulated inspired by the tracking model predictive control framework, and it considers the central idea of including artificial variables into the control problem to obtain a closed-loop system with an enlarged domain of attraction and with feasibility insurance. Finally, numerical results in a case study considering a quadrotor UAV are provided to corroborate the proposed strategy. © 2021 IEEE.

关键词： Energy efficiency

来源：评论

学校读者我要写书评

暂无评论

Image-based Korean Cattle Weight Estimation System using Top View Image

引用

Transactions of the Korean Institute of Electrical Engineers 2022年第1期71卷 246-252页

作者： Lee, Chang-Bok Park, Yeon-Soo Cho, Hyun-Chong Dept. of Electronics Engineering and Interdisciplinary Graduate Program for BIT Medical Convergence Kangwon National University Korea Republic of Dept. of Interdisciplinary Graduate Program for BIT Medical Convergence Kangwon National University Korea Republic of Gangwon-do Livestock Research Institute Korea Republic of

Weight information is important to understand the growth of cattle and to calculate an appropriate daily feed amount. However, since the method of measuring weight using a scale consumes a lot of labor and time, weight information is not actively used in Korean cattle farms. Image-based weight estimation can be a new weight estimation method that does not cause stress to cattle and does not require labor. In this study, top-view data that is easy to apply to actual Korean cattle farms was used. A segmentation mask was obtained by dividing the cow and the background, and weight was estimated by extracting weight correlation factors from the segmentation mask. The weight estimation of the method of proposing area, body length, and body width, which are verified factors based on related studies, reached an error rate of 8.5%, and when the extracted grid elements were added to improve performance, it reached 7.5%. © 2022 Korean Institute of Electrical Engineers. All rights reserved.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Assessment of force control improvement induced by sinusoidal vibrotactile stimulation in dominant and non-dominant hands

引用

Research on Biomedical engineering 2021年第1期37卷 95-103页

作者： Germer, Carina Marconi Moreira, Luciana Sobral Elias, Leonardo Abdala Department of Electronics and Biomedical Engineering School of Electrical and Computer Engineering University of Campinas CampinasSP Brazil Neural Engineering Research Laboratory Center for Biomedical Engineering University of Campinas CampinasSP Brazil Cellular and Structural Biology Graduate Program Institute of Biology University of Campinas CampinasSP Brazil Faculty of Biological Sciences Center for Life Sciences Pontifical Catholic University of Campinas CampinasSP Brazil

Purpose: Vibrotactile stimulation has been shown to improve sensorimotor integration in the central nervous system, thereby enhancing motor performance in several conditions. Here we aim to address two questions: (1) Would lateral asymmetries yielding handedness influence the force control improvement induced by sinusoidal vibrotactile stimulation? (2) Does the force control improvement depend on finding an optimum stimulus intensity? Methods: Eleven participants performed a visually guided force-matching task, while different intensities of sinusoidal vibrations were applied to the index finger skin. Tasks consisted of abductions of the index fingers of dominant and non-dominant hands at 5% of their maximal voluntary contractions. The force coefficient of variation (CoV) in the condition without vibration was used as the reference (control) for later comparison with five vibration intensities. Also, force CoV in the control condition was compared with those obtained from the optimal vibration selected either from both hands or from a single hand (the one that started the experiment). Results: Despite the known sensory asymmetries in cortical and spinal circuits, handedness did not significantly influence the improvement of force control induced by sinusoidal vibrotactile stimulation during low-intensity contractions. Moreover, there was a significant reduction in force variability irrespective of the criteria used to select the vibration intensity. Conclusion: These findings add to the current knowledge on the effects of sinusoidal vibrotactile stimulation on force control and have potential implications for the development of wearable motor enhancer devices. © 2020, Sociedade Brasileira de Engenharia Biomedica.

关键词： Force control

来源：评论

学校读者我要写书评

暂无评论

BAQET: BRAM-aware Quantization for Efficient Transformer Inference via Stream-based Architecture on an FPGA 25

BAQET: BRAM-aware Quantization for Efficient Transformer Inf...

引用

Proceedings of the 2025 ACM/SIGDA International Symposium on Field programmable Gate Arrays

作者： Ling-Chi Yang Chi-Jui Chen Trung Le Bo-Cheng Lai Scott Hauck Shih-Chieh Hsu Institute of Electronics National Yang Ming Chiao Tung University Hsinchu Taiwan Graduate Degree Program of College of Electrical and Computer Engineering in National Yang Ming Chiao Tung University Hsinchu Taiwan Electrical and Computer Engineering in University of Washington Seattle USA Electrical and Computer Engineering University of Washington Seattle USA Physics University of Washington Seattle USA

ISBN: (纸本)9798400713965

FPGAs are a compelling substrate for supporting machine learning inference. Tools such as High-Level Synthesis and hls4ml can shorten the development cycle for deploying ML algorithms on FPGAs, but can struggle to handle the large on-chip storage needed for many of these models. In particular the high BRAM usage found in many of these flows can cause Place & Route failures during synthesis. In this paper we propose using a Simulated-Annealing based flow to perform BRAM-aware quantization. This approach trades off inference accuracy with BRAM usage, to provide a high-quality inference engine that still meets on-chip resource constraints. We demonstrate this flow for Transformer-based machine learning algorithms, which include Flash Attention in a Stream-based Dataflow architecture. Our system imposes minimal accuracy drops, yet can reduce BRAM usage by 20%-50%, and improve power efficiency by 264%-812% compared to existing Transformer-based accelerators on FPGAs

关键词： efficient transformer

来源：评论

学校读者我要写书评

暂无评论

Reachability Analysis of Nonlinear Discrete-Time Systems Using Polyhedral Relaxations and Constrained Zonotopes

arXiv

引用

arXiv 2025年

作者： Rego, Brenner S. Raffo, Guilherme V. Terra, Marco H. Scott, Joseph K. Department of Chemical and Biomolecular Engineering Georgia Institute of Technology 311 Ferst Dr. AtlantaGA30318 United States Department of Electrical and Computer Engineering University of São Paulo SP São Carlos13566-590 Brazil Department of Electronics Engineering the Graduate Program in Electrical Engineering Federal University of Minas Gerais MG Belo Horizonte31270-901 Brazil

This paper presents a novel algorithm for reachability analysis of nonlinear discrete-time systems. The proposed method combines constrained zonotopes (CZs) with polyhedral relaxations of factorable representations of nonlinear functions to propagate CZs through nonlinear functions, which is normally done using conservative linearization techniques. The new propagation method provides better approximations than those resulting from linearization procedures, leading to significant improvements in the computation of reachable sets in comparison to other CZ methods from the literature. Numerical examples highlight the advantages of the proposed algorithm. © 2025, CC BY-NC-ND.

关键词： Numerical methods

来源：评论

学校读者我要写书评

暂无评论

Predicting Microsatellite Instability from Histology Images with Dilated Neighborhood Attention Transformer in Colorectal Cancer

Predicting Microsatellite Instability from Histology Images ...

引用

IEEE Symposium on Bioinformatics and Bioengineering (BIBE)

作者： Yun-Chi Chen Tung Chao Wimaya Nitya Phandita Tzu-Yu Sun Huai-Zhi Wang Yi-Hsien Hsieh Le-Yin Hsu Che Lin Graduate Institute of Biomedical Electronics and Bioinformatics National Taiwan University (NTU) Taipei Taiwan College of Bioresources and Agriculture NTU Taipei Taiwan Department of Computer Science and Information Engineering National Taiwan University of Science and Technology Taipei Taiwan Graduate Institute of Communication Engineering NTU Taipei Taiwan Data Science Degree Program NTU Taipei Taiwan Department of Electrical Engineering NTU Taipei Taiwan Center for Advanced Computing and Imaging in Biomedicine NTU Taipei Taiwan Smart Medicine and Health Informatics Program NTU Taipei Taiwan

ISBN: (数字)9798331518622

ISBN: (纸本)9798331518639

Microsatellite instability (MSI) is a pivotal genetic marker influencing the efficacy of immunotherapy in colorectal cancer. Traditional MSI examination often requires additional genetic or immunohistochemical tests, whereas histology images, widely available in colorectal cancer diagnosis, offer a valuable alternative for MSI prediction. Although Transformer-based models have demonstrated promising outcomes in predicting MSI from histology images, they are hampered by traditional local attention mechanisms that struggle to capture long-range interdependencies and establish a comprehensive global receptive field. In this study, we introduce DiNAT-MSI, a novel framework for histology-based MSI prediction that incorporates the Dilated Neighborhood Attention Transformer (DiNAT). This model enhances global context recognition and substantially expands receptive fields, all without additional computational burden. Our results demonstrate that DiNAT-MSI achieves a superior patientwise AUROC compared to ResNet18 and Swin Transformer, along with commendable explainability. Our work not only illustrates a more accessible diagnostic tool for leveraging histological data but also underscores the potential of Transformerbased models with sophisticated attention designs in advancing precision medicine for colorectal cancer patients.

关键词： Attention mechanisms Histopathology Computational modeling Precision medicine Small satellites Immunotherapy Colorectal cancer Predictive models Transformers Genetics

来源：评论

学校读者我要写书评

暂无评论

Three-phase Self-Excited Induction Generator Operating as Single-phase Induction Generator using Static VAR Compensator

Three-phase Self-Excited Induction Generator Operating as Si...

引用

International Conference on Electrical engineering/electronics, Computer, Telecommunications and Information Technology, ECTI-CON

作者： Sawat Yukhalang Theppanom Sopapirm Vijit Kinnares The Electrical Engineering Graduate Program Faculty of Engineering and Technology Mahanakorn University of Technology Nongjok Bangkok Thailand Institute of Electrical and Electronics Faculty of Engineering and TechnologyMahanakorn University of Technology Nongjok Bangkok Thailand Department of Electrical Engineering School of Engineering King Mongkut’s Institute Technology Ladkrabang Chalongkrung Road Ladkrabang Bangkok Thailand

This paper proposes a performance analysis of using a static volt-amp-reactive (VAR) compensator for a three-phase self-excited induction generator (SEIG) operating as a single-phase induction generator. The study proposes appropriate capacitor for building up voltage supplying power to load with static VAR for regulating terminal voltage. The simulation model is analyzed by using MATLAB/Simulink under various conditions such as reactive power supplied, dynamic response, steady state, building up voltage for startup, linear load, non-linear load and harmonics in system. From the study (based on the study, the findings revealed it is found) that using the static VAR compensator connected in parallel with loads for supplying reactive power to the load systems is needed for maintaining terminal voltage.

关键词：

来源：评论

学校读者我要写书评

暂无评论

NFT-Based Hydroponic Automated Control Using Adaptive Network-Based Fuzzy Inference System

NFT-Based Hydroponic Automated Control Using Adaptive Networ...

引用

Robotics, Automation and Artificial Intelligence (RAAI), IEEE International Conference on

作者： Nico Surantha Vito Vincentdo BINUS Graduate Program - Master of Computer Science Bina Nusantara University Jakarta Indonesia Department of Electrical Faculty of Scienc and EngineeringElectronics and Communication Engineering Tokyo City University Japan

The rapid population growth and industrial development in developing countries harm the agricultural sector because many agricultural lands are converted into residential or industrial areas. Applying modern agriculture technologies such as hydroponic could help to overcome the problem. However, hydroponics requires special attention in adjusting the pH and nutrient levels to maximize plant growth, so an automated system is needed to manage the process. In this research, a smart hydroponic system is proposed by applying Adaptive Networkbased Fuzzy Inference System (ANFIS) and Internet-of-Things. The IoT system consists of sensor, actuator, and data processing layer is designed to monitor and control the condition of pH and nutrition of the observed plants. Then, the ANFIS algorithm is designed to control the level of pH and nutrition. The experiment results show that the system can automatically adjust the pH and nutrient levels to the expected range for growing plants, and the fuzzy controller made using ANFIS are more accurate and stable than the fuzzy controller made using Sugeno. This study shows that ANFIS has excellent performance when controlling multiple actuators, as long as the data set has great granularity and well defined.

关键词： Fuzzy logic Training Actuators Adaptive systems Hydroponics Media Control systems

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：