检索结果-内蒙古大学图书馆

Dental Detection and Classification of YOLOv3-SPP based on Convolutional Block Attention Module

学校读者我要写书评

暂无评论

Dental Detection and Classification of YOLOv3-SPP based on C...

International Conference on Computer and Communications (ICCC)

作者： Ning Li Ruifeng Guo Xiaozhou Liu Lin Wu Hongliang Wang Shenyang Institute of Computing Technology Chinese Academy of Sciences Shenyang China University of Chinese Academy of Sciences Shenyang China Liaoning Provience Human-Computer Interaction System Engineering Research Center Based on Digital Twin Shenyang China Liaoning Provincial Key Laboratory of Oral Diseases School and Hospital of Stomatology China Medical University Shenyang China

The purpose of this research paper is to implement the tooth target detection task by deep convolutional neural networks. In order to solve the problems of low accuracy of target detection due to the high similarity between teeth and complex tooth textures, an improved tooth detection method with YOLOv3-SPP model is proposed in this paper. This method incorporates the convolutional block attention module (CBAM) in the YOLOv3-SPP algorithm framework, and improves the performance of the network features by adding the channel attention mechanism and spatial attention mechanism to the feature extraction network to enhance the saliency of the tooth target region in the image. Secondly, CIoU border regression loss is introduced to improve the localization accuracy. In addition, a Non-maximum suppression (NMS) method is used to solve the candidate frame overlap problem. Through experiments, it is shown that the mAP of the modified YOLOv3-SPP target detection model is improved to 86.8 percent, which indicates that the improved model can be applied to the detection of teeth.

关键词： Location awareness Training Teeth Object detection Feature extraction Dentistry Classification algorithms

Emotion Recognition Empowered Human-Computer Interaction with Domain Adaptation Network

学校读者我要写书评

暂无评论

IEEE Transactions on Consumer Electronics 2024年

作者： Xu, Xu Fu, Chong Chen, Junxin Northeastern University School of Computer Science and Engineering Shenyang110004 China Dalian University of Technology China The School of Software Dalian116621 China Northeastern University School of Computer Science and Engineering Shenyang110004 China Northeastern University Key Laboratory of Intelligent Computing in Medical Image Ministry of Education Shenyang110819 China Ministry of Education Engineering Research Center of Security Technology of Complex Network System Shenyang110004 China Dalian University of Technology School of Software Dalian116621 China

Multi-modal emotion recognition plays a vital role in the human-computer interaction (HCI) for consumer electronics. Nowadays, many studies have developed multi-modal fusion algorithms for this ***, two challenging issues remain unsolved, i.e., inefficient multi-modal feature fusion and unclear distance in feature *** this end, we develop a novel framework, namely LAFDA-Net, for cross-subject emotion recognition using EEG and eye movement *** is based on low-rank fusion and domain adaptation *** specifically, the multi-modal signals are input into the feature extraction branch in parallel to generate ***, these features are fused by the low-rank fusion branch, reducing complexity and avoiding ***, the fused features are flattened and sent to the classification branch to determine the emotion *** training, these features are input into the domain adaptive branch to bridge the gap between the source domain and target *** benchmark datasets, i.e., SEED, SEED-IV, and SEED-V, are employed for performance *** results demonstrate that the proposed LAFDA-Net is robust, effective, and has advantages over peer methods. © 1975-2011 IEEE.

关键词： Emotion Recognition

Privileged Anatomical and Protocol Discrimination in Trackerless 3D Ultrasound Reconstruction

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Qi Shen, Ziyi Li, Qian Barratt, Dean C. Dowrick, Thomas Clarkson, Matthew J. Vercauteren, Tom Hu, Yipeng Centre for Medical Image Computing Wellcome/EPSRC Centre for Interventional and Surgical Sciences Department of Medical Physics and Biomedical Engineering University College London London United Kingdom State Key Laboratory of Robotics and System Harbin Institute of Technology Harbin China School of Biomedical Engineering & Imaging Sciences King’s College London London United Kingdom

Three-dimensional (3D) freehand ultrasound (US) reconstruction without using any additional external tracking device has seen recent advances with deep neural networks (DNNs). In this paper, we first investigated two identified contributing factors of the learned inter-frame correlation that enable the DNN-based reconstruction: anatomy and protocol. We propose to incorporate the ability to represent these two factors - readily available during training - as the privileged information to improve existing DNN-based methods. This is implemented in a new multitask method, where the anatomical and protocol discrimination are used as auxiliary tasks. We further develop a differentiable network architecture to optimise the branching location of these auxiliary tasks, which controls the ratio between shared and task-specific network parameters, for maximising the benefits from the two auxiliary tasks. Experimental results, on a dataset with 38 forearms of 19 volunteers acquired with 6 different scanning protocols, show that 1) both anatomical and protocol variances are enabling factors for DNN-based US reconstruction;2) learning how to discriminate different subjects (anatomical variance) and predefined types of scanning paths (protocol variance) both significantly improve frame prediction accuracy, volume reconstruction overlap, accumulated tracking error and final drift, using the proposed algorithm. © 2023, CC BY.

关键词： Deep neural networks

Trackerless freehand ultrasound with sequence modelling and auxiliary transformation over past and future frames

学校读者我要写书评

暂无评论

arXiv 2022年

Three-dimensional (3D) freehand ultrasound (US) reconstruction without a tracker can be advantageous over its two-dimensional or tracked counterparts in many clinical applications. In this paper, we propose to estimate 3D spatial transformation between US frames from both past and future 2D images, using feed-forward and recurrent neural networks (RNNs). With the temporally available frames, a further multitask learning algorithm is proposed to utilise a large number of auxiliary transformation-predicting tasks between them. Using more than 40,000 US frames acquired from 228 scans on 38 forearms of 19 volunteers in a volunteer study, the hold-out test performance is quantified by frame prediction accuracy, volume reconstruction overlap, accumulated tracking error and final drift, based on ground-truth from an optical tracker. The results show the importance of modelling the temporal-spatially correlated input frames as well as output transformations, with further improvement owing to additional past and/or future frames. The best performing model was associated with predicting transformation between moderately-spaced frames, with an interval of less than ten frames at 20 frames per second (fps). Little benefit was observed by adding frames more than one second away from the predicted transformation, with or without LSTM-based RNNs. Interestingly, with the proposed approach, explicit within-sequence loss that encourages consistency in composing transformations or minimises accumulated error may no longer be required. The implementation code and volunteer data1 will be made publicly available ensuring reproducibility and further research. © 2022, CC BY.

关键词： Ultrasonics

Joint object contour points and semantics for instance segmentation

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhang, Wenchao Fu, Chong Zhu, Mai Cao, Lin Tie, Ming Sham, Chiu-Wing School of Computer Science and Engineering Northeastern University Shenyang110819 China Engineering Research Center of Security Technology of Complex Network System Ministry of Education China Key Laboratory of Intelligent Computing in Medical Image Ministry of Education Northeastern University Shenyang110819 China School of Information and Communication Engineering Beijing Information Science and Technology University Beijing100101 China Science and Technology on Space Physics Laboratory Beijing100076 China School of Computer Science University of Auckland New Zealand

The attributes of object contours has great significance for instance segmentation task. However, most of the current popular deep neural networks do not pay much attention to the object edge information. Inspired by the human annotation process when making instance segmentation datasets, in this paper, we propose Mask Point R-CNN aiming at promoting the neural network’s attention to the object boundary. Specifically, we innovatively extend the original human keypoint detection task to the contour point detection of any object. Based on this analogy, we present an contour point detection auxiliary task to Mask R-CNN, which can boost the gradient flow between different tasks by effectively using feature fusion strategies and multi-task joint training. As a consequence, the model will be more sensitive to the edges of the object and can capture more geometric features. Quantitatively, the experimental results show that our approach outperforms vanilla Mask R-CNN by 3.8% on Cityscapes dataset and 0.8% on COCO dataset. Copyright © 2020, The Authors. All rights reserved.

关键词：

BuildSenSys: Reusing building sensing data for traffic prediction with cross-domain learning

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Fan, Xiaochen Xiang, Chaocan Chen, Chao Yang, Panlong Gong, Liangyi Song, Xudong Nanda, Priyadarsi He, Xiangjian School of Electrical and Data Engineering Faculty of Engineering and Information Technology University of Technology SydneyNSW2007 Australia Key Laboratory of Dependable Service Computing in Cyber Physical Society Chongqing University Ministry of Education China College of Computer Science Chongqing University Chongqing400044 China School of Computer Science and Technology University of Science and Technology of China Hefei Anhui230026 China School of Software BNRist Tsinghua University Beijing100084 China

With the rapid development of smart cities, smart buildings are generating a massive amount of building sensing data by the equipped sensors. Indeed, building sensing data provides a promising way to enrich a series of data-demanding and cost-expensive urban mobile applications. In this paper, as a preliminary exploration, we study how to reuse building sensing data to predict traffic volume on nearby roads. Compared with existing studies, reusing building sensing data has considerable merits of cost-efficiency and high-reliability. Nevertheless, it is non-trivial to achieve accurate prediction on such cross-domain data with two major challenges. First, relationships between building sensing data and traffic data are not unknown as prior, and the spatio-temporal complexities impose more difficulties to uncover the underlying reasons behind the above relationships. Second, it is even more daunting to accurately predict traffic volume with dynamic building-traffic correlations, which are cross-domain, non-linear, and time-varying. To address the above challenges, we design and implement BuildSenSys, a first-of-its-kind system for nearby traffic volume prediction by reusing building sensing data. Our work consists of two parts, i.e., Correlation Analysis and Cross-domain Learning. First, we conduct a comprehensive building-traffic analysis based on multi-source datasets, disclosing how and why building sensing data is correlated with nearby traffic volume. Second, we propose a novel recurrent neural network for traffic volume prediction based on cross-domain learning with two attention mechanisms. Specifically, a cross-domain attention mechanism captures the building-traffic correlations and adaptively extracts the most relevant building sensing data at each predicting step. Then, a temporal attention mechanism is employed to model the temporal dependencies of data across historical time intervals. The extensive experimental studies demonstrate that BuildSenSys outperf

关键词： Internet of things

Research on adaptive strategy for end-hopping system

学校读者我要写书评

暂无评论

Tongxin Xuebao/Journal on Communications 2011年第11 A期32卷 51-57页

作者： Zhao, Chun-Lei Jia, Chun-Fu Weng, Chen Lin, Kai College of Information Technical Science Nankai University Tianjin 300071 China Key Laboratory of Computer Vision and System Ministry of Education Tianjin University of Technology Tianjin 300384 China Tianjin Key Lab. of Intelligent Computing and Novel Software Technology Tianjin University of Technology Tianjin 300384 China

An adaptive strategy was proposed for EH system. Modules of attack detection, feedback transmission and adaptive control were added to original EH system. They could help to make threats analysis and adjust hopping strategy in real time, so as to reduce the probability of the system stepping into the attackers' target area. With the experiments proving the anti-attack ability of the prototype system, the feasibility and effectiveness of adaptive strategy were demonstrated for EH system.

关键词： Network security

Using Data-Level Parallelism to Accelerate Instruction -Level Redundancy

学校读者我要写书评

暂无评论

Using Data-Level Parallelism to Accelerate Instruction -Leve...

World Automation Congress

作者： Yu Hu Zhong liang Chen Xiaowei Li Key Laboratory of Computer System and Architecture Institute of Computing Technology CAS Beijing 100190 China Department of Electrical and Computer Engineering Northeastern University Boston MA 02115 USA

ISBN: (纸本)9781467344975

Instruction-level redundancy is an effective scheme to reduce the susceptibility of microprocessors to soft errors, offering high error detection and recovery capability;however, it usually incurs significant performance degradation due to resource racing. Motivated by the fact that narrow-width operands are commonly seen in applications, we exploit data-level parallelism to accelerate instruction-level redundancy. For the instructions within sphere of replication (SoR) of data-level redundancy, normal and redundant versions of the narrow-width operand of the instruction are folded into one register to share the same functional unit during execution hence alleviating resource racing. The other instructions are all protected by instructionlevel redundancy. We run SPECint2000 benchmarks on a modified version of SimpleScalar simulator, and synthesize the extra hardware to evaluate area overhead of the proposed pipeline. Experimental results show that our acceleration scheme outperforms conventional instruction-level redundancy by 13% in IPC. Besides, the extra area overhead is negligible.

关键词： keywords Data-level redundancy Instruction-level redundancy Narrow-width value Sphere of replication

Gaussian Decay Function-based Improved Moment Matching for Ocular Artifacts Removal

学校读者我要写书评

暂无评论

Gaussian Decay Function-based Improved Moment Matching for O...

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Hua Jiang Qiuxia Shi Siying Liu Jiuying Zhang Qinglin Zhao Bin Hu Gansu Provincial Key Laboratory of Wearable Computing School of Information Science and Engineering Key Laboratory of Special Functional Materials and Structural Design Ministry of Education Lanzhou University Lanzhou China CAS Center for Excellence in Brain Science and Intelligence Technology Shanghai Institutes for Biological Sciences Chinese Academy of Sciences Chinese Academy of Sciences Joint Research Center for Cognitive Neurosensor Technology of Lanzhou University & Institute of Semiconductors Engineering Research Center of Open Source Software and Real-Time System (Lanzhou University) Ministry of Education Key Laboratory of Special Functional Materials and Structural Design (Lanzhou University) Ministry of Education Lanzhou China

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

Electroencephalogram (EEG) equipped with high time resolution that distracted by incoherent brain sources which including ocular artifacts (OAs) generating by blinks is of great significance. Therefore, these OAs got corrected before extracting information from EEG is indispensable. Improved Moment Matching (IMM) is a high-speed denoising algorithm suitable for removing OA in multi-channel EEG, which is an improvement of the moment matching method used to remove stripe noise in hyperspectral images. On foundation of this, this paper proposes an optimization algorithm for IMM based on Gaussian decay function (IMM_G). In the first place, the construction of the reference signal is optimized via utilizing a Gaussian decay function, thereby preventing the signal distortion caused by the filter. And then, a method for judging the to-be-processed interval is proposed to realizes the individualized processing of different channels. As a result, through quantitative comparison experiments with simulated data and real data from the UAIS laboratory, it was identified that IMM_G showed less time complexity, significantly enhanced arithmetic speed and denoising consequence while the detail retention ability of the original method for the non-blinking region of multi-channel EEG data is maintained. Hence, this method could be extensively used in High-density EEG (hdEEG) OAs removal scene.

关键词： Electrooculography Noise reduction Filtering algorithms Distortion Electroencephalography Recording Time-domain analysis