检索结果-内蒙古大学图书馆

International Conference on Computer Vision (ICCV)

作者： Ke Xu Lei Han Ye Tian Shangshang Yang Xingyi Zhang School of Artificial Intelligence Anhui University Hefei China Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Hefei China School of Computer Science and Technology Anhui University Hefei China Institutes of Physical Science and Information Technology Anhui University Hefei China

Current model quantization methods have shown their promising capability in reducing storage space and computation complexity. However, due to the diversity of quantization forms supported by different hardware, one limitation of existing solutions is that usually require repeated optimization for different scenarios. How to construct a model with flexible quantization forms has been less studied. In this paper, we explore a one-shot network quantization regime, named Elastic Quantization Neural Networks (EQ-Net), which aims to train a robust weight-sharing quantization supernet. First of all, we propose an elastic quantization space (including elastic bit-width, granularity, and symmetry) to adapt to various mainstream quantitative forms. Secondly, we propose the Weight Distribution Regularization Loss (WDR-Loss) and Group Progressive Guidance Loss (GPG-Loss) to bridge the inconsistency of the distribution for weights and output logits in the elastic quantization space gap. Lastly, we incorporate genetic algorithms and the proposed Conditional Quantization-Aware Accuracy Predictor (CQAP) as an estimator to quickly search mixed-precision quantized neural networks in supernet. Extensive experiments demonstrate that our EQ-Net is close to or even better than its static counterparts as well as state-of-the-art robust bit-width methods. Code can be available at https://***/xuke225/EQ-Net.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Adversarial Learning Network for Recovering Rppg signals from Low-Resolution Images for Remote Heart Rate Measurement

SSRN

引用

SSRN 2024年

作者： Zhai, Dezhao Chen, Wei Li, Qinwei Yu, Ming Wu, Hang Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin University of Technology Tianjin300384 China Systems Engineering Institute Academy of Military Sciences People's Liberation Army Tianjin300161 China Tianjin Key Laboratory for Advanced Signal Processing Civil Aviation University of China Tianjin300300 China School of Artificial Intelligence Nankai University Tianjin300381 China

Remote Photoplethysmography (rPPG) is a non-contact method for measuring heart rate (HR) through facial video, which has significant research value and application potential. However, in long-distance vital sign measurement tasks, the low resolution of facial images and insufficient pixels in the facial region severely degrade the quality of the rPPG signals, leading to a sharp decline in HR accuracy. To address this issue, this paper proposes an adversarial learning network for recovering rPPG signals from low-resolution images (ALRSL-HR). First, arbitrary resolution face images are fed into multi-channel network to extract rPPG signal features capturing rich rPPG information, generating rPPG feature blocks, and temporally aggregating for richer feature representation. These features are then inputted into the backbone network to generate adversarial examples of rPPG signals of varying qualities. Finally, a two-stage adversarial loss function is employed to enable the low-quality rPPG signals to learn the periodic characteristics of high-quality rPPG signals, recover the low-quality signals, and narrow the quality gap between rPPG signals from low-resolution images. Extensive experiments were conducted in this paper to validate the effectiveness of our model on the public COHFACE, UBFC-Phys and UBFC-rPPG datasets. The results of the experiments demonstrate that the proposed ALRSL-HR model can effectively recover degraded rPPG signals from low-resolution images. © 2024, The Authors. All rights reserved.

关键词： Photoplethysmography

来源：评论

学校读者我要写书评

暂无评论

Federated Client-tailored Adapter for Medical Image Segmentation

arXiv

引用

arXiv 2025年

作者： Hu, Guyue Song, Siyuan Kang, Yukun Yin, Zhu Zhao, Gangming Li, Chenglong Tang, Jin Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Anhui Provincial Key Laboratory of Security Artificial Intelligence School of Artificial Intelligence Anhui University Hefei230601 China School of Internet Anhui University Hefei230601 China Key Laboratory of Intelligent Computing and Signal Processing of Ministry of Education Anhui Provincial Key Laboratory of Multimodal Cognitive Computation School of Computer Science and Technology Anhui University Hefei230601 China Department of Computer Science The University of Hong Kong Hong Kong

Medical image segmentation in X-ray images is beneficial for computer-aided diagnosis and lesion localization. Existing methods mainly fall into a centralized learning paradigm, which is inapplicable in the practical medical scenario that only has access to distributed data islands. Federated Learning has the potential to offer a distributed solution but struggles with heavy training instability due to client-wise domain heterogeneity (including distribution diversity and class imbalance). In this paper, we propose a novel Federated Client-tailored Adapter (FCA) framework for medical image segmentation, which achieves stable and client-tailored adaptive segmentation without sharing sensitive local data. Specifically, the federated adapter stirs universal knowledge in off-the-shelf medical foundation models to stabilize the federated training process. In addition, we develop two client-tailored federated updating strategies that adaptively decompose the adapter into common and individual components, then globally and independently update the parameter groups associated with common client-invariant and individual client-specific units, respectively. They further stabilize the heterogeneous federated learning process and realize optimal client-tailored instead of sub-optimal global-compromised segmentation models. Extensive experiments on three large-scale datasets demonstrate the effectiveness and superiority of the proposed FCA framework for federated medical segmentation. © 2025, CC BY.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Challenge-Aware RGBT Tracking 1

引用

16th European Conference on Computer Vision, ECCV 2020

作者： Li, Chenglong Liu, Lei Lu, Andong Ji, Qing Tang, Jin Key Lab of Intelligent Computing and Signal Processing of Ministry of Education Anhui Provincial Key Laboratory of Multimodal Cognitive Computation School of Computer Science and Technology Anhui University Hefei230601 China

ISBN: (数字)9783030585426

ISBN: (纸本)9783030585419

RGB and thermal source data suffer from both shared and specific challenges, and how to explore and exploit them plays a critical role to represent the target appearance in RGBT tracking. In this paper, we propose a novel challenge-aware neural network to handle the modality-shared challenges (e.g., fast motion, scale variation and occlusion) and the modality-specific ones (e.g., illumination variation and thermal crossover) for RGBT tracking. In particular, we design several parameter-shared branches in each layer to model the target appearance under the modality-shared challenges, and several parameter-independent branches under the modality-specific ones. Based on the observation that the modality-specific cues of different modalities usually contains the complementary advantages, we propose a guidance module to transfer discriminative features from one modality to another one, which could enhance the discriminative ability of some weak modality. Moreover, all branches are aggregated together in an adaptive manner and parallel embedded in the backbone network to efficiently form more discriminative target representations. These challenge-aware branches are able to model the target appearance under certain challenges so that the target representations can be learnt by a few parameters even in the situation of insufficient training data. From the experimental results we will show that our method operates at a real-time speed while performing well against the state-of-the-art methods on three benchmark datasets. © 2020, Springer Nature Switzerland AG.

关键词： Computers

来源：评论

学校读者我要写书评

暂无评论

Hand Gesture Recognition Using Surface Electromyography and 3d Self-Organizing Map Network with Switching States

SSRN

引用

SSRN 2024年

作者： Duan, Yahan Li, Wenyu Solé-Casals, Jordi Liu, Zhongxin Duan, Feng College of Artificial Intelligence Nankai University 38 Tongyan Road Jinnan District Tianjin300350 China Tianjin Key Laboratory of Interventional Brain-Computer Interface and Intelligent Rehabilitation 38 Tongyan Road Jinnan District Tianjin300350 China Data and Signal Processing Group University of Vic Central University of Catalonia Vic Catalonia08500 Spain

Surface electromyography (sEMG) is widely utilized in rehabilitation therapy. However, existing gesture recognition methods predominantly concentrate on rigid static gestures, overlooking the importance of dynamic switching in practical tasks and the associated dynamic features for natural hand control. This neglect limits the consideration of hand gesture dexterity and diversity. In response, this work proposes a novel paradigm for hand gesture recognition, in which the switching state between two gestures is treated equally as a static pattern for downstream feature extraction and classification. A 3-dimensional self-organizing mapping neural network is investigated to fully characterize the high-dimensional sEMG signals. Due to the sequential feature of hand gesture variation, the prior switching probability is further applied to reduce ambiguity and computational consumption. The proposed paradigm and algorithm are tested and verified through five-fold cross-validation with multiple subjects. The experimental results show that the average accuracy of the 3D self-organizing mapping method is 78.74% while considering switching states, which is at least 6.97% higher than some traditional learning-based algorithms: long short-term memory and wavelet neural network. This work suggests the potential for achieving dexterous control in prosthetics. © 2024, The Authors. All rights reserved.

关键词： Gesture recognition

来源：评论

学校读者我要写书评

暂无评论

Automated Detection of Epileptic Spikes and Seizures Incorporating a Novel Spatial Clustering Prior

arXiv

引用

arXiv 2025年

作者： Dong, Hanyang Sheng, Shurong Wang, Xiongfei Gao, Jia-Hong Sun, Yi Yang, Wanli Xiao, Kuntao Teng, Pengfei Luan, Guoming Lv, Zhao School of Artificial Intelligence Anhui University Hefei230601 China Anhui Province Key Laboratory of Biomedical Imaging and Intelligent Processing Institute of Artificial Intelligence Hefei Comprehensive National Science Center Hefei230088 China Department of Neurosurgery Beijing Key Laboratory of Epilepsy Sanbo Brain Hospital Capital Medical University Beijing China McGovern Institute for Brain Research Peking University Beijing China Center for MRI Research Academy for Advanced Interdisciplinary Studies Peking University Beijing China Beijing City Key Lab for Medical Physics and Engineering Institute of Heavy Ion Physics School of Physics Peking University Beijing China Anhui Province Key Laboratory of Multimodal Cognitive Computation School of Computer Science and Technology Anhui University Hefei230601 China

A Magnetoencephalography (MEG) time-series recording consists of multi-channel signals collected by superconducting sensors, with each signal’s intensity reflecting magnetic field changes over time at the sensor location. Automating epileptic MEG spike detection significantly reduces manual assessment time and effort, yielding substantial clinical benefits. Existing research addresses MEG spike detection by encoding neural network inputs with signals from all channel within a time segment, followed by classification. However, these methods overlook simultaneous spiking occurred from nearby sensors. We introduce a simple yet effective paradigm that first clusters MEG channels based on their sensor’s spatial position. Next, a novel convolutional input module is designed to integrate the spatial clustering and temporal changes of the signals. This module is fed into a custom MEEG-ResNet3D developed by the authors, which learns to extract relevant features and classify the input as a spike clip or not. Our method achieves an F1 score of 94.73% on a large real-world MEG dataset Sanbo-CMR collected from two centers, outperforming state-of-the-art approaches by 1.85%. Moreover, it demonstrates efficacy and stability in the Electroencephalographic (EEG) seizure detection task, yielding an improved weighted F1 score of 1.4% compared to current state-of-the-art techniques evaluated on TUSZ, whch is the largest EEG seizure dataset. Copyright © 2025, The Authors. All rights reserved.

关键词： Magnetoencephalography

来源：评论

学校读者我要写书评

暂无评论

MSCA-UNet: Multi-Scale Channel Adaptive Sensing Module for Medical Image Segmentation

MSCA-UNet: Multi-Scale Channel Adaptive Sensing Module for M...

引用

New Trends in computational intelligence (NTCI), International Conference on

作者： Liangjin Diao Longfeng Shen Jiacong Chen Fangzhen Ge Mengwen Li Anhui Engineering Research Center for Intelligent Computing and Application on Cognitive Behavior (ICACB) College of Computer Science and Technology Huaibei Normal University Huaibei Anhui China Institute of Artificial Intelligence Hefei Comprehensive National Science Center Hefei China Key Laboratory of Intelligent Computing & Signal Processing Ministry of Education Anhui University Hefei Anhui China

ISBN: (数字)9798331517021

ISBN: (纸本)9798331517038

Recent advancements in CNNs for medical image segmentation have focused on the combination of CNN and Transformer architectures to capture both local and global features. However, challenges remain, such as effectively utilizing fused features at the encoder-decoder bottleneck and addressing blurred object edges in datasets. To tackle these issues, we introduce a multi-scale channel adaptive sensing module, enhancing the model's representational and perceptual capabilities by integrating multi-scale and channel-specific information. Additionally, we propose a cross-perception module and multi-scale segmentation head to improve global coordination and multi-scale information gathering. Experiments on the ACDC and Synapse datasets demonstrate our model's superiority.

关键词： Image segmentation Adaptation models Head Image edge detection Computer architecture Transformers Market research Sensors Synapses Biomedical imaging

来源：评论

学校读者我要写书评

暂无评论

Multiscale Low-Frequency Memory Network for Improved Feature Extraction in Convolutional Neural Networks

arXiv

引用

arXiv 2024年

作者： Wu, Fuzhi Wu, Jiasong Kong, Youyong Yang, Chunfeng Yang, Guanyu Shu, Huazhong Carrault, Guy Senhadji, Lotfi Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications Southeast University Bangladesh Laboratoire Traitement du Signal et de l'Image Univ Rennes France Centre de Recherche en Information Biomédicale Sino-français CRIBs France Jiangsu Provincial Joint International Research Laboratory of Medical Information Processing Southeast University Bangladesh

Deep learning and Convolutional Neural Networks (CNNs) have driven major transformations in diverse research areas. However, their limitations in handling low-frequency information present obstacles in certain tasks like interpreting global structures or managing smooth transition images. Despite the promising performance of transformer structures in numerous tasks, their intricate optimization complexities highlight the persistent need for refined CNN enhancements using limited resources. Responding to these complexities, we introduce a novel framework, the Multiscale Low-Frequency Memory (MLFM) Network, with the goal to harness the full potential of CNNs while keeping their complexity unchanged. The MLFM efficiently preserves low-frequency information, enhancing performance in targeted computer vision tasks. Central to our MLFM is the Low-Frequency Memory Unit (LFMU), which stores various low-frequency data and forms a parallel channel to the core network. A key advantage of MLFM is its seamless compatibility with various prevalent networks, requiring no alterations to their original core structure. Testing on ImageNet demonstrated substantial accuracy improvements in multiple 2D CNNs, including ResNet, MobileNet, EfficientNet, and ConvNeXt. Furthermore, we showcase MLFM's versatility beyond traditional image classification by successfully integrating it into image-to-image translation tasks, specifically in semantic segmentation networks like FCN and U-Net. In conclusion, our work signifies a pivotal stride in the journey of optimizing the efficacy and efficiency of CNNs with limited resources. This research builds upon the existing CNN foundations and paves the way for future advancements in computer vision. Our codes are available at https://***/AlphaWuSeu/MLFM. © 2024, CC BY-NC-ND.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

DL2G: Anatomical Landmark Detection with Deep Local Features and Geometric Global Constraint

DL2G: Anatomical Landmark Detection with Deep Local Features...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Rui Wang Wanli Yang Kuntao Xiao Yi Sun Shurong Sheng Zhao Lv Jiahong Gao Auhui University Hefei China Anhui Province Key Laboratory of Biomedical Imaging and Intelligent Processing Institute of Artificial Intelligence Hefei Comprehensive National Science Center Hefei China Anhui Province Key Laboratory of Multimodal Cognitive Computation Hefei China Center for MRI Research Academy for Advanced Interdisciplinary Studies Peking University Beijing China Beijing City Key Lab for Medical Physics and Engineering Institute of Heavy Ion PhysicsSchool of Physics Peking University Beijing China McGovern Institute for BrainResearch Peking University Beijing China

ISBN: (数字)9798350386226

ISBN: (纸本)9798350386233

Anatomical landmark detection, a pivotal research area in medical image processing, holds immense value in surgical navigation, image registration, and related fields. Traditional machine learning methods struggle with generalization and robustness. Current supervised end-to-end approaches lacks in-terpretability in capturing global information, and GPU memory constraints restrict their application to extensive 3D medical image datasets. Moreover, existing methods overlook intrinsic geometric cues within images and point sets. Herein, we introduce a novel anatomical landmark detection framework that integrates deep learning’s representation capabilities with global geometric information derived from images and point sets (DL2G). This approach facilitates anatomical landmark localization in a local-to-global fashion. Initially, we train a deep feature descriptor based on self-supervised contrastive learning, which is further used to screen candidate points for the target one by comparing local patch embeddings. Subsequently, geometric constraints constructed from labeled template point sets enable the selection of geometrically consistent matching point sets. Ultimately, the framework utilizes the global mutual information of images to perform landmark localization through iterative optimization. Adhering to the AFID data annotation protocol, we evaluated our method on two public MRI head datasets, OASIS and HCP, for detecting 32 anatomical landmarks. The results convincingly demonstrate the superiority of our method over current state-of-the-art approaches in terms of both accuracy and computational efficiency.

关键词： Location awareness Three-dimensional displays Protocols Navigation Memory management Surgery Feature extraction Robustness Optimization Mutual information

来源：评论

学校读者我要写书评

暂无评论

SC-RegNet: Unsupervised Self-Calibrated 3D Biomedical Image Registration Network

SC-RegNet: Unsupervised Self-Calibrated 3D Biomedical Image ...

引用

Chinese Automation Congress (CAC)

作者： Jian Yang Jun Wu Rui Zhang Meng Wang Yujie Tang Lei Qu Key Laboratory of Computational Intelligence and Signal Processing Ministry of Education School of Electronics and Information Engineering Anhui University (AHU) Hefei China

ISBN: (数字)9781665426473

ISBN: (纸本)9781665426480

Deformable image registration is critical in many biomedical image analysis tasks. Traditional registration algorithms involve a time-consuming iterative optimization process. Most deep learning-based registration networks follow the U-Net shape encoding-decoding architecture, and their downsampling layers are shallow, making it difficult to obtain sufficient semantic information with a small receptive field. In this work, we extend the self-calibrated convolution applied to tasks such as 2D target detection to the encoding stage of 3D image registration networks. The proposed SC-RegNet can get more semantic information to make up for the existing registration networks in an unsupervised training manner while not introducing additional parameters. We evaluated SC-RegNet on two public available human Brain MRI datasets: LPBA40 and IBSR18, and a private mouse brain dataset. The experimental results show the superiority of SC-RegNet over the SOTA image registration methods with no extra time-consuming.

关键词： Training Image registration Three-dimensional displays Convolution Shape Semantics Object detection

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：