检索结果-内蒙古大学图书馆

FedATA: Adaptive attention aggregation for federated self-supervised medical image segmentation

NEUROCOMPUTING 2025年 613卷

作者： Dai, Jian Wu, Hao Liu, Huan Yu, Liheng Hu, Xing Liu, Xiao Geng, Daoying Fudan Univ Acad Engn & Technol Shanghai 200433 Peoples R China Fudan Univ Huashan Hosp Dept Radiol Shanghai 200040 Peoples R China Fudan Univ Huashan Hosp Dept Dermatol Shanghai 200040 Peoples R China Univ Shanghai Sci & Technol Sch Opt Elect & Comp Engn Shanghai 20093 Peoples R China

Pre-trained on large-scale datasets has profoundly promoted the development of deep learning models in medical image analysis. For medical image segmentation, collecting a large number of labeled volumetric medical images from multiple institutions is an enormous challenge due to privacy concerns. Self-supervised learning with mask image modeling (MIM) can learn general representation without annotations. Integrating MIM into FL enables collaborative learning of an efficient pre-trained model from unlabeled data, followed by fine-tuning with limited annotations. However, setting pixels as reconstruction targets in traditional MIM fails to facilitate robust representation learning due to the medical image's complexity and distinct characteristics. On the other hand, the generalization of the aggregated model in FL is also impaired under the heterogeneous data distributions among institutions. To address these issues, we proposed a novel self-supervised federated learning, which combines masked self-distillation with adaptive attention federated learning. Such incorporation enjoys two vital benefits. First, masked self-distillation sets high-quality latent representations of masked tokens as the target, improving the descriptive capability of the learned presentation rather than reconstructing low-level pixels. Second, adaptive attention aggregation with Personalized federate learning effectively captures specific-related representation from the aggregated model, thus facilitating local fine-tuning performance for target tasks. We conducted comprehensive experiments on two medical segmentation tasks using a large-scale dataset consisting of volumetric medical images from multiple institutions, demonstrating superior performance compared to existing federated self-supervised learning approaches.

关键词： Self-supervised learning masked image modeling Federated learning

来源：评论

学校读者我要写书评

暂无评论

A Unified Self-Supervised Learning Framework for Hyperspectral image Classification

引用

IEEE ACCESS 2025年 13卷 49874-49890页

作者： Zhang, Guoyun Tan, Ping Fang, Xiaoqi Zhou, Xuhui Zhao, Lin Hunan Inst Sci & Technol Yueyang 414000 Peoples R China

Self-supervised learning (SSL) methods, including contrastive learning (CL) and masked image modeling (MIM), have shown commendable performance on various remote sensing visual tasks. However, learning comprehensive and complementary representations from unlabeled remote sensing images remains a challenging task due to their intricate spatial-spectral characteristics. To address this issue, we propose a Unified Self-supervised Learning Framework (USeLF) that incorporates both CL and MIM to enhance hyperspectral image (HSI) classification. USeLF employs an asymmetric Siamese network framework, designed to accommodate both CL and MIM. Leveraging the common characteristics of these two paradigms, we developed a hierarchical Transformer structure that integrates downsampling and upsampling modules to improve model generalization. Furthermore, image masking is devised as a novel data augmentation strategy to bridge the gap between CL and MIM. USeLF learns diverse and complementary features from both CL and MIM, resulting in superior performance for HSI classification. Experiments conducted on four widely used HSI datasets demonstrate that USeLF outperforms existing state-of-the-art (SOTA) methods. Notably, on the WHU-Hi-HongHu dataset-a complex agricultural scene with numerous crop classes-USeLF achieves a 5.3% improvement in overall accuracy compared to the current SOTA method.

关键词： contrastive learning masked image modeling hyperspectral image classification Self-supervised learning Self-supervised learning contrastive learning masked image modeling hyperspectral image classification

来源：评论

学校读者我要写书评

暂无评论

MIFAE-Forensics: masked image-Frequency AutoEncoder for DeepFake Detection

MIFAE-Forensics: Masked Image-Frequency AutoEncoder for Deep...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Wang, Hanyi Liu, Zihan Wang, Shilin School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University China

ISBN: (纸本)9798350368741

With continuously evolving generative models and increasingly diverse face forgery products, there is a growing demand for DeepFake detectors with stronger generalization ability and robustness. Previous works mainly capture method-specific forgery artifacts in the training set, thus failing to generalize well to unseen manipulations. In this paper, our key insight is that exploring common characteristics of natural faces is more ideal to alleviate overfitting rather than relying on specific forgery clues, as all sorts of manipulated images have intrinsic distributional differences from those captured by cameras. Hence, we propose a two-stage method, termed MIFAE-Forensics. Specifically, it reconstructs both facial semantics and local details from masked facial regions and high-frequency components, respectively, aiming to capture natural facial consistency in spatial domain and high-frequency details in frequency domain simultaneously. This facilitates the learning of a robust and transferable facial representation specialized for DeepFake detection. Subsequently, the pre-trained model is further fine-tuned to perform binary forgery classification along with reconstructing real faces in spatial domain, which ensures that the detector can maintain the ability to model real faces and encourages it to make decisions based on reconstruction discrepancies. Extensive experiments show superior results over state-of-the-arts on a wide range of DeepFake detection benchmarks. Our code is available at https://***/Mark-Dou/Forensics. © 2025 IEEE.

关键词： DeepFake Detection masked image modeling

来源：评论

学校读者我要写书评

暂无评论

BossNAS Family: Block-Wisely Self-Supervised Neural Architecture Search

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2025年第5期47卷 3500-3514页

作者： Li, Changlin Lin, Sihao Tang, Tao Wang, Guangrun Li, Mingjie Liang, Xiaodan Chang, Xiaojun Univ Technol Sydney Australian Artificial Intelligence Inst Broadway NSW 2007 Australia RMIT Univ Sch Comp Technol Melbourne Vic 3000 Australia Sun Yat Sen Univ Guangzhou 510275 Peoples R China Stanford Univ Stanford CA 94305 USA Univ Sci & Technol China Sch Informat Sci & Technol Future Media Comp Lab Hefei 230052 Peoples R China Mohamed Bin Zayed Univ Artificial Intelligence MBZ Dept Comp Vis Abu Dhabi U Arab Emirates

Recent advances in hand-crafted neural architectures for visual recognition underscore the pressing need to explore architecture designs comprising diverse building blocks. Concurrently, neural architecture search (NAS) methods have gained traction as a means to alleviate human efforts. Nevertheless, the question of whether NAS methods can efficiently and effectively manage diversified search spaces featuring disparate candidates, such as Convolutional Neural Networks (CNNs) and transformers, remains an open question. In this work, we introduce a novel unsupervised NAS approach called BossNAS (Block-wisely Self-supervised Neural Architecture Search), which aims to address the problem of inaccurate predictive architecture ranking caused by a large weight-sharing space while mitigating potential ranking issue caused by biased supervision. To achieve this, we factorize the search space into blocks and introduce a novel self-supervised training scheme called Ensemble Bootstrapping, to train each block separately in an unsupervised manner. In the search phase, we propose an unsupervised Population-Centric Search, optimizing the candidate architecture towards the population center. Additionally, we enhance our NAS method by integrating masked image modeling and present BossNAS++ to overcome the lack of dense supervision in our block-wise self-supervised NAS. In BossNAS++, we introduce the training technique named masked Ensemble Bootstrapping for block-wise supernet, accompanied by a masked Population-Centric Search scheme to promote fairer architecture selection. Our family of models, discovered through BossNAS and BossNAS++, delivers impressive results across various search spaces and datasets. Our transformer model discovered by BossNAS++ attains a remarkable accuracy of 83.2% on imageNet with only 10.5B MAdds, surpassing DeiT-B by 1.4% while maintaining a lower computation cost. Moreover, our approach excels in architecture rating accuracy, achieving Spearman correlat

关键词： Computer architecture Transformers Training Computational modeling Accuracy Visualization Correlation Architecture Computer vision Neural architecture search Neural architecture search (NAS) vision transformer masked image modeling self-supervised learning unsupervised NAS

来源：评论

学校读者我要写书评

暂无评论

Personvit: large-scale self-supervised vision transformer for person re-identification

引用

MACHINE VISION AND APPLICATIONS 2025年第2期36卷 1-13页

作者： Hu, Bin Wang, Xinggang Liu, Wenyu Huazhong Univ Sci & Technol Inst Artificial Intelligence 1037 Luoyu Rd Wuhan 430074 Hubei Peoples R China Huazhong Univ Sci & Technol Sch EIC 1037 Luoyu Rd Wuhan 430074 Hubei Peoples R China

Person Re-Identification (ReID) aims to retrieve relevant individuals in non-overlapping camera images and has a wide range of applications in the field of public safety. In recent years, with the development of Vision Transformer (ViT) and self-supervised learning techniques, the performance of person ReID based on self-supervised pre-training has been greatly improved. Person ReID requires extracting highly discriminative local fine-grained features of the human body, while traditional ViT is good at extracting context-related global features, making it difficult to focus on local human body features. To this end, this article introduces the recently emerged masked image modeling (MIM) self-supervised learning method into person ReID, and effectively extracts high-quality global and local features through large-scale unsupervised pre-training by combining masked image modeling and discriminative contrastive learning, and then conducts supervised fine-tuning training in the person ReID task. This person feature extraction method based on ViT with masked image modeling (PersonViT) has the good characteristics of unsupervised, scalable, and strong generalization capabilities, overcoming the problem of difficult annotation in supervised person ReID, and achieves state-of-the-art results on publicly available benchmark datasets, including MSMT17, Market1501, DukeMTMC-reID, and Occluded-Duke. The code and pre-trained models of the PersonViT method are released at https://***/hustvl/PersonViT to promote further research in the person ReID field.

关键词： Person ReID Self-supervised learning masked image modeling ViT

来源：评论

学校读者我要写书评

暂无评论

Frequency-Domain Masking and Spatial Interaction for Generalizable Deepfake Detection

引用

ELECTRONICS 2025年第7期14卷 1302-1302页

作者： Luo, Xinyu Wang, Yu Jilin Univ JLU Coll Comp Sci & Technol Qianjin St 2699 Changchun 130012 Peoples R China Jilin Univ Key Lab Symbol Computat & Knowledge Engn Minist Educ Qianjin St 2699 Changchun 130012 Peoples R China

Over the past few years, the rapid development of deepfake technology based on generative models has posed a significant threat to the field of information security. Despite the notable progress in deepfake-detection methods based on the spatial domain, the detection capability of the models drops sharply when dealing with low-quality images. Moreover, the effectiveness of detection relies on the realism of the forged images and the specific traces inherent to particular forgery techniques, which often weakens the models' generalization ability. To address this issue, we propose the Frequency-Domain Masking and Spatial Interaction (FMSI) model. The FMSI model innovatively introduces masked image modeling in frequency-domain processing. This prevents the model from focusing too much on specific frequency-domain features and enhances its generalization ability. We design a high-frequency information convolution module for spatial and channel dimensions to help the model capture subtle forgery traces more effectively. Also, we creatively design a dual stream architecture for frequency-domain and spatial-domain information interaction and overcome single-domain detection limitations. Our model is tested on three public benchmark datasets (FaceForensics++, Celeb-DF, and WildDeepfake) through intra-domain and cross-domain experiments. The detection and generalization capabilities of the model are evaluated using the AUC and EER metrics. The experimental results demonstrate that our model not only possesses high detection capability but also exhibits excellent generalization ability.

关键词： deepfake detection masked image modeling Fourier transform vision transformer

来源：评论

学校读者我要写书评

暂无评论

SARATR-X: Toward Building a Foundation Model for SAR Target Recognition

引用

IEEE TRANSACTIONS ON image PROCESSING 2025年 34卷 869-884页

作者： Li, Weijie Yang, Wei Hou, Yuenan Liu, Li Liu, Yongxiang Li, Xiang Natl Univ Def Technol Coll Elect Sci & Technol Changsha 410073 Peoples R China Shanghai AI Lab Shanghai 200000 Peoples R China

Despite the remarkable progress in synthetic aperture radar automatic target recognition (SAR ATR), recent efforts have concentrated on detecting and classifying a specific category, e.g., vehicles, ships, airplanes, or buildings. One of the fundamental limitations of the top-performing SAR ATR methods is that the learning paradigm is supervised, task-specific, limited-category, closed-world learning, which depends on massive amounts of accurately annotated samples that are expensively labeled by expert SAR analysts and have limited generalization capability and scalability. In this work, we make the first attempt towards building a foundation model for SAR ATR, termed SARATR-X. SARATR-X learns generalizable representations via self-supervised learning (SSL) and provides a cornerstone for label-efficient model adaptation to generic SAR target detection and classification tasks. Specifically, SARATR-X is trained on 0.18 M unlabelled SAR target samples, which are curated by combining contemporary benchmarks and constitute the largest publicly available dataset till now. Considering the characteristics of SAR images, a backbone tailored for SAR ATR is carefully designed, and a two-step SSL method endowed with multi-scale gradient features was applied to ensure the feature diversity and model scalability of SARATR-X. The capabilities of SARATR-X are evaluated on classification under few-shot and robustness settings and detection across various categories and scenes, and impressive performance is achieved, often competitive with or even superior to prior fully supervised, semi-supervised, or self-supervised algorithms. Our SARATR-X and the curated dataset are released at https://***/waterdisappear/SARATR-X to foster research into foundation models for SAR image interpretation.

关键词： Radar polarimetry Foundation models Remote sensing Imaging Frequency modulation Buildings Object detection Noise Marine vehicles Atmospheric modeling Synthetic aperture radar target recognition object detection foundation model self-supervised learning deep learning masked image modeling

来源：评论

学校读者我要写书评

暂无评论

Beyond masking: Demystifying token-based pre-training for vision transformers

引用

PATTERN RECOGNITION 2025年 162卷

作者： Tian, Yunjie Xie, Lingxi Fang, Jiemin Jiao, Jianbin Tian, Qi Univ Chinese Acad Sci Beijing 100049 Peoples R China Huawei Inc Beijing Peoples R China Huazhong Univ Sci & Technol Wuhan Peoples R China

masked image modeling (MIM) has rapidly advanced over the past year, primarily using vision transformers to learn self-supervised visual representations by masking parts of an input image and recovering the missing contents. While MIM has shown promising results on downstream tasks, we explore whether alternative methods can also effectively "learn by recovering missing contents". In this paper, we design six learning objectives that degrade images in different ways, including zooming-in, zooming-out, distorting, shuffling, blurring, and de-colorizing. Through extensive experiments, we identify key design principles for token-based pre-training of vision transformers. Notably, the best results are achieved by preserving the original image style and incorporating spatial misalignment into masking. For example, combining masking with zooming-in achieves superior performance over MIM on challenging tasks with negligible additional cost, highlighting the potential for more effective self-supervised pre-training strategies.

关键词： Self-supervised learning Vision transformers Token-based pre-training masked image modeling

来源：评论

学校读者我要写书评

暂无评论

Evolved Hierarchical Masking for Self-Supervised Learning

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2025年第2期47卷 1013-1027页

作者： Feng, Zhanzhou Zhang, Shiliang Peking Univ Sch Comp Sci State Key Lab Multimedia Informat Proc Beijing 100871 Peoples R China Peng Cheng Lab Shenzhen 518055 Peoples R China

Existing masked image modeling methods apply fixed mask patterns to guide the self-supervised training. As those mask patterns resort to different criteria to depict image contents, sticking to a fixed pattern leads to a limited vision cues modeling capability. This paper introduces an evolved hierarchical masking method to pursue general visual cues modeling in self-supervised learning. The proposed method leverages the vision model being trained to parse the input visual cues into a hierarchy structure, which is hence adopted to generate masks accordingly. The accuracy of hierarchy is on par with the capability of the model being trained, leading to evolved mask patterns at different training stages. Initially, generated masks focus on low-level visual cues to grasp basic textures, then gradually evolve to depict higher-level cues to reinforce the learning of more complicated object semantics and contexts. Our method does not require extra pre-trained models or annotations and ensures training efficiency by evolving the training difficulty. We conduct extensive experiments on seven downstream tasks including partial-duplicate image retrieval relying on low-level details, as well as image classification and semantic segmentation that require semantic parsing capability. Experimental results demonstrate that it substantially boosts performance across these tasks. For instance, it surpasses the recent MAE by 1.1% in imageNet-1K classification and 1.4% in ADE20K segmentation with the same training epochs. We also align the proposed method with the current research focus on LLMs. The proposed approach bridges the gap with large-scale pre-training on semantic demanding tasks and enhances intricate detail perception in tasks requiring low-level feature recognition.

关键词： Visualization Training Semantics Self-supervised learning Semantic segmentation image classification Computational modeling Neurons Representation learning Predictive models masked image modeling efficient learning model pretraining

来源：评论

学校读者我要写书评

暂无评论

Centroid-Centered modeling for Efficient Vision Transformer Pre-Training 7th

Centroid-Centered Modeling for Efficient Vision Transformer ...

引用

7th Chinese Conference on Pattern Recognition and Computer Vision

作者： Van, Xin Li, Zuchao Zhang, Lefei Wuhan Univ Wuhan Peoples R China

ISBN: (纸本)9789819785049;9789819785056

masked image modeling (MIM) is a new self-supervised vision pre-training paradigm using Vision Transformer (ViT). Previous works can be pixel-based or token-based, using original pixels or discrete visual tokens from parametric tokenizer models, respectively. Our proposed centroid-based approach, CCViT, leverages k-means clustering to obtain centroids for image modeling without supervised training of the tokenizer model, which only takes seconds to create. This non-parametric centroid tokenizer only takes seconds to create and is faster for token inference. The centroids can represent both patch pixels and index tokens with the property of local invariance. Specifically, we adopt patch masking and centroid replacing strategies to construct corrupted inputs, and two stacked encoder blocks to predict corrupted patch tokens and reconstruct original patch pixels. Experiments show that our CCViT achieves 84.4% top-1 accuracy on imageNet-1K classification with ViT-B and 86.0% with ViT-L. We also transfer our pre-trained model to other downstream tasks. Our approach achieves competitive results with recent baselines without external supervision and distillation training from other models.

关键词： Large-scale visual pre-training Self-supervised learning masked image modeling

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：