检索结果-内蒙古大学图书馆

26th International Conference on Medical image Computing and Computer-Assisted Intervention (MICCAI)

作者： Pan, Jiazhen Shit, Suprosanna Turgut, Oezguen Huang, Wenqi Li, Hongwei Bran Stolt-Anso, Nil Kuestner, Thomas Hammernik, Kerstin Rueckert, Daniel Tech Univ Munich Klinikum Rechts Isar Sch Med Munich Germany Univ Zurich Dept Quantitat Biomed Zurich Switzerland Tech Univ Munich Sch Comp Informat & Technol Munich Germany Univ Tubingen Hosp Med Image & Data Anal Tubingen Germany Imperial Coll London Dept Comp London England

ISBN: (纸本)9783031439988;9783031439995

In dynamic Magnetic Resonance Imaging (MRI), k-space is typically undersampled due to limited scan time, resulting in aliasing artifacts in the image domain. Hence, dynamic MR reconstruction requires not only modeling spatial frequency components in the x and y directions of k-space but also considering temporal redundancy. Most previous works rely on image-domain regularizers (priors) to conduct MR reconstruction. In contrast, we focus on interpolating the undersampled k-space before obtaining images with Fourier transform. In this work, we connect masked image modeling with k-space interpolation and propose a novel Transformer-based k-space Global Interpolation Network, termed k-GIN. Our k-GIN learns global dependencies among low- and high-frequency components of 2D+t k-space and uses it to interpolate unsampled data. Further, we propose a novel k-space Iterative Refinement Module (k-IRM) to enhance the high-frequency components learning. We evaluate our approach on 92 in-house 2D+t cardiac MR subjects and compare it to MR reconstruction methods with image-domain regularizers. Experiments show that our proposed k-space interpolation method quantitatively and qualitatively outperforms baseline methods. Importantly, the proposed approach achieves substantially higher robustness and generalizability in cases of highly-undersampled MR data.

关键词： Cardiac MR Imaging Reconstruction k-space Interpolation masked image modeling masked Autoencoders Transformers

来源：评论

学校读者我要写书评

暂无评论

CXRMIM: masked image modeling PRE-TRAINING PARADIGM FOR CHEST X-RAY imageS ANALYSIS 30

CXRMIM: MASKED IMAGE MODELING PRE-TRAINING PARADIGM FOR CHES...

引用

30th IEEE International Conference on image Processing (ICIP)

作者： Wang, Zhendong Ma, Haowen Niu, Jianwei Beihang Univ Sino French Engn Sch Beijing 100191 Peoples R China Yizhun Med AI Co Ltd Beijing 100088 Peoples R China Beihang Univ Sch Comp Sci & Engn Beijing 100191 Peoples R China

ISBN: (纸本)9781728198354

As an effective approach for Vision Transformers (ViT) to obtain better initializations and representations in natural image analysis, masked image modeling (MIM), performs the pretext task of reconstructing images by adopting partial observations without any label. Several works adopted dissimilar mask strategies to make ViT aggregate contextual information to infer missed contents. Nonetheless, chest radiographs conspicuously differ from photographic images, and conducting MIM in chest X-rays remains challenging. On that account, this paper came up with a specialized pre-training recipe cxr-MIM and a masking strategy for chest radiographs on the basis of their physiological characters. In cxrMIM, the out-of-lung region was first analyzed, and the lung region was then reconstructed with the help of mechanical connections and similarities of anatomy and physiology. cxrMIM facilitates ViT to excavate the commonalities of pulmonary structures and promote better performance on downstream tasks. We conducted experiments on the ChestX-ray 14 dataset using advanced self-supervised methods (e.g. MoCo v3, MAE) for comparison. Quantitative and qualitative results signified that cxrMIM reinforced the efficiency of Vision Transformer to resolve the multi-label thorax disease classification problem, and cxrMIM pretrained Swin-B performed comparably to the state-of-the-art CNN models.

关键词： Chest X-ray images analysis masked image modeling self-supervised pre-training

来源：评论

学校读者我要写书评

暂无评论

Exploring masked image modeling for Face Anti-spoofing 5th

Exploring Masked Image Modeling for Face Anti-spoofing

引用

5th Chinese Conference on Pattern Recognition and Computer Vision (PRCV)

作者： Ma, Xuetao Zhang, Jun Zhang, Yunfei Zhou, Daoxiang Taiyuan Univ Technol Coll Data Sci Taiyuan 030024 Peoples R China

ISBN: (纸本)9783031189067;9783031189074

Face anti-spoofing (FAS) is an indispensable step in face recognition systems. In order to distinguish spoofing faces from genuine ones, existing methods always require sophisticated handcrafted features or well-designed supervised networks to learn discriminative representation. In this paper, a novel generative self-supervised learning inspired FAS approach is proposed, which has three merits: no need for massive labeled images, excellent discriminative ability, and the learned features have good transferability. Firstly, in the pretext task, the masked image modeling strategy is exploited to learn general fine-grained features via image patches reconstruction in an unsupervised encoder-decoder structure. Secondly, the encoder knowledge is transferred into the downstream FAS task. Finally, the entire network parameters are fine-tuned using only binary labels. Extensive experiments on three standard benchmarks demonstrate that our method can be exceedingly close to the state-of-the-art in FAS, which indicates that masked image modeling is able to learn discriminative face detail features that are beneficial to FAS.

关键词： Face anti-spoofing Self-supervised learning masked image modeling Transformer network

来源：评论

学校读者我要写书评

暂无评论

Pre-Training For mmWave Radar Object Detection Through masked image modeling 2

Pre-Training For mmWave Radar Object Detection Through Maske...

引用

2nd International Conference on image Processing, Computer Vision and Machine Learning, ICICML 2023

作者： Zhuang, Long Jiang, Tiezhen School of Integrated Circuits Anhui University Hefei China

ISBN: (纸本)9798350331417

Millimeter-wave (mmWave) radar object detection (ROD) is an indispensable technique in autonomous driving technology because of its ability to maintain stability in complex environments. Deep learning (DL) techniques have made promising achievements in radar target detection techniques, but these techniques usually require large labeled datasets. However, the labeling of millimeter-wave radar datasets is more complex. In this paper, we introduce a novel pre-training method based on mask image modeling (MIM) in the ROD task, aiming to address the dataset limitations in the development of ROD techniques. The method performs spatial and temporal masking operations on unlabeled millimeter-wave radar images, which can acquire more spatial and temporal features simultaneously in the pre-training stage. Furthermore, inspired by T-RODNet, this paper presents a lightweight SS-RODNet network for testing the effectiveness of the proposed MIM method. The model aims to fully integrate the advantages of extracting local and global features. Numerous ablation experiments prove that the proposed MIM method is effective, and SS-RODNet achieves second place on the CRUW datasets. © 2023 IEEE.

关键词： 3D CNN autonomous driving masked image modeling Radar object detection Transformer

来源：评论

学校读者我要写书评

暂无评论

Intra-modality masked image modeling: A self-supervised pre-training method for brain tumor segmentation

引用

BIOMEDICAL SIGNAL PROCESSING AND CONTROL 2024年第PartA期95卷

作者： Qi, Liangce Shi, Weili Miao, Yu Li, Yonghui Feng, Guanyuan Jiang, Zhengang Changchun Univ Sci & Technol Dept Comp Sci & Technol Changchun 130022 Jilin Peoples R China

Despite the great success of deep neural networks in brain tumor segmentation, it is challenging to obtain sufficient annotated images due to the requirement of clinical expertise. masked image modeling recently achieved competitive performance compared with supervised training by learning rich representations from unlabeled data. However, it is originally designed for vision transformers and its effectiveness has not been well -studied in the medical domain, usually for limited unlabeled data and small convolutional network scenarios. In this paper, we propose a self -supervised learning framework to pre -train U -Net for brain tumor segmentation. Our goal is to learn modality -specific and modality -invariant representations from multimodality magnetic resonance images. This is motivated by the fact that different modalities indicate the same organs and tissues but have various appearances. To achieve this, we design a new pretext task that reconstructs the masked patches of each modality based on the partial observation of other modalities. We evaluate our method by transfer performance on BraTS 2020 dataset. The experimental results demonstrate our method outperforms other self -supervised learning methods and improves the performance of a strong fully supervised baseline. The source codes are available at https://***/mobiletomb/IS-MIM.

关键词： masked image modeling Self-supervised learning Brain tumor segmentation

来源：评论

学校读者我要写书评

暂无评论

A self-supervised image aesthetic assessment combining masked image modeling and contrastive learning

引用

JOURNAL OF VISUAL COMMUNICATION AND image REPRESENTATION 2024年 101卷

作者： Yang, Shuai Wang, Zibei Wang, Guangao Ke, Yongzhen Qin, Fan Guo, Jing Chen, Liming Tiangong Univ Sch Comp Sci & Technol Tianjin Peoples R China Tianjin Key Lab Autonomous Intelligence Technol & Tianjin Peoples R China Nankai Univ Business Sch Tianjin Peoples R China Tiangong Univ Natl Demonstrat Ctr Expt Engn Training Educ Tianjin Peoples R China Fitow Tianjin Detect Technol Co LTD Tianjin Peoples R China

Learning more abundant image features helps improve the image aesthetic assessment task performance. masked image modeling (MIM) is implemented based on the Vision Transformer (ViT), which learns pixel-level features while reconstructing images. Contrastive learning pulls in the same image features while pushing away different image features in the feature space to learn high-level semantic features. Since contrastive learning and MIM capture different levels of image features, combining these two methods could learn more rich feature representations and thus promote the performance of aesthetic assessment. Therefore, we propose a pretext task combining contrastive learning and MIM with learning richer image features. In this approach, the original image is randomly masked and reconstructed on the online network. The reconstructed and original images composition the positive example to calculate the contrastive loss on the target network. In the experiment on the AVA dataset, our method obtained better performance than the baseline.

关键词： image aesthetic assessment Self-supervised learning masked image modeling

来源：评论

学校读者我要写书评

暂无评论

High-Quality Object Detection Method for UAV images Based on Improved DINO and masked image modeling

引用

REMOTE SENSING 2023年第19期15卷 4740-4740页

作者： Lu, Wanjie Niu, Chaoyang Lan, Chaozhen Liu, Wei Wang, Shiju Yu, Junming Hu, Tao PLA Strateg Support Force Informat Engn Univ Inst Data & Target Engn Zhengzhou 450052 Peoples R China China Elect Technol Grp Corp Res Inst 27 Zhengzhou 450047 Peoples R China

The extensive application of unmanned aerial vehicle (UAV) technology has increased academic interest in object detection algorithms for UAV images. Nevertheless, these algorithms present issues such as low accuracy, inadequate stability, and insufficient pre-training model utilization. Therefore, a high-quality object detection method based on a performance-improved object detection baseline and pretraining algorithm is proposed. To fully extract global and local feature information, a hybrid backbone based on the combination of convolutional neural network (CNN) and vision transformer (ViT) is constructed using an excellent object detection method as the baseline network for feature extraction. This backbone is then combined with a more stable and generalizable optimizer to obtain high-quality object detection results. Because the domain gap between natural and UAV aerial photography scenes hinders the application of mainstream pre-training models to downstream UAV image object detection tasks, this study applies the masked image modeling (MIM) method to aerospace remote sensing datasets with a lower volume than mainstream natural scene datasets to produce a pre-training model for the proposed method and further improve UAV image object detection accuracy. Experimental results for two UAV imagery datasets show that the proposed method achieves better object detection performance compared to state-of-the-art (SOTA) methods with fewer pre-training datasets and parameters.

关键词： UAV image object detection masked image modeling global-local hybrid

来源：评论

学校读者我要写书评

暂无评论

Deblurring masked image modeling for ultrasound image analysis

引用

MEDICAL image ANALYSIS 2024年 97卷 103256页

作者： Kang, Qingbo Lao, Qicheng Gao, Jun Liu, Jingyan Yi, Huahui Ma, Buyun Zhang, Xiaofan Li, Kang Sichuan Univ West China Hosp Dept Ultrasonog Chengdu 610041 Sichuan Peoples R China Sichuan Univ West China Hosp West China Biomed Big Data Ctr Chengdu 610041 Sichuan Peoples R China Beijing Univ Posts & Telecommun Sch Artificial Intelligence Beijing 100876 Peoples R China Shanghai Artificial Intelligence Lab Shanghai 200030 Peoples R China Sichuan Univ Coll Comp Sci Chengdu 610041 Sichuan Peoples R China Shanghai Jiao Tong Univ Shanghai 200240 Peoples R China

Recently, large pretrained vision foundation models based on masked image modeling (MIM) have attracted unprecedented attention and achieved remarkable performance across various tasks. However, the study of MIM for ultrasound imaging remains relatively unexplored, and most importantly, current MIM approaches fail to account for the gap between natural images and ultrasound, as well as the intrinsic imaging characteristics of the ultrasound modality, such as the high noise-to-signal ratio. In this paper, motivated by the unique high noise-to-signal ratio property in ultrasound, we propose a deblurring MIM approach specialized to ultrasound, which incorporates a deblurring task into the pretraining proxy task. The incorporation of deblurring facilitates the pretraining to better recover the subtle details within ultrasound images that are vital for subsequent downstream analysis. Furthermore, we employ a multi-scale hierarchical encoder to extract both local and global contextual cues for improved performance, especially on pixel-wise tasks such as segmentation. We conduct extensive experiments involving 280,000 ultrasound images for the pretraining and evaluate the downstream transfer performance of the pretrained model on various disease diagnoses (nodule, Hashimoto's thyroiditis) and task types (classification, segmentation). The experimental results demonstrate the efficacy of the proposed deblurring MIM, achieving state-of-the-art performance across a wide range of downstream tasks and datasets. Overall, our work highlights the potential of deblurring MIM for ultrasound image analysis, presenting an ultrasound-specific vision foundation model.

关键词： image deblurring masked image modeling Self-supervised learning Ultrasound image analysis Vision foundation model

来源：评论

学校读者我要写书评

暂无评论

Mask autoencoder for enhanced image reconstruction with position coding offset and combined masking

引用

VISUAL COMPUTER 2025年 1-15页

作者： Wang, Yuenan Wang, Hua Zhang, Fan Shandong Technol & Business Univ Sch Informat & Elect Engn Yantai 264005 Shandong Peoples R China Ludong Univ Sch Informat & Elect Engn Yantai 264025 Shandong Peoples R China Shandong Technol & Business Univ Sch Comp Sci & Technol Yantai 264005 Shandong Peoples R China Shandong Technol & Business Univ Shandong Future Intelligent Financial Engn Lab Yantai 264005 Shandong Peoples R China

Existing masked image modeling (MIM) methods mainly fill and enhance images by modeling and filling masked areas. These techniques usually borrow principles from computer graphics, where similar methods are used to synthesize missing parts of images or meshes. However, current MIM methods still have problems such as incomplete capture of local information and failure to capture dynamic information. To address the above situation, we propose a multi-mask autoencoder (M-MAE). M-MAE borrows the smooth transition technology from computer graphics, combines patch masking and random masking, and enhances the stability of the model by optimizing the processing of masked areas during training. In addition, we introduce a scaling layer (LScale) to improve training dynamics, similar to the scaling and distortion transformations used in graphics to adapt to different spatial distributions. To further improve the accuracy of the model in capturing spatial relationships, we propose a position encoding offset method that can generate more spatially aware encodings, thereby effectively enhancing the spatial expression ability of the model. Comparative experiments show that M-MAE can achieve an Acc1 of 84.5 on the imageNet-1K dataset, an Acc1 of 58.2 on the Places365 dataset, and an mIoU of 49.4 on the ADE20K dataset. From the experimental results, it can be seen that our innovation improves the performance of the model in multiple visual tasks and provides a tighter integration between visual and graphics methods. Code is released at (https://***/zoomba35/mmae).

关键词： masked image modeling masked autoencoder Vision transformer Position encoding Computer graphics

来源：评论

学校读者我要写书评

暂无评论

Self-supervised deep hash image deduplication algorithm for visual inspection of concrete

引用

JOURNAL OF CIVIL STRUCTURAL HEALTH MONITORING 2025年 1-18页

作者： Du, Hao Wang, Huifeng Jiao, Yunmei Shan, Yuanhe He, Zhihui Zhang, Xiaowei Pan, Zefeng Gao, Rong Huang, He Changan Univ Sch Elect & Control Engn Xian 710064 Peoples R China Univ Shaanxi Prov Engn Res Ctr Vehicle Rd Integrat Percept & Intelli Xian 710064 Peoples R China

Concrete visual inspection is crucial for assessing its safety condition. Aiming at the issue of excessive overlap and redundancy of images collected by UAVs or wall-climbing robots during inspection, this research proposed an image deduplication algorithm based on self-supervised deep hashing, where the images are generated into binary hash codes of concrete apparent multi-scale features by SwinFpnHashCLR model, and efficiently deduplicated between the images is realized by the hamming distance and Otsu's algorithm. The experiment results indicated that the algorithm proposed in this paper achieved a deduplication rate of 47.73% on the concrete surface dataset, and its effect and robustness were better than the other method. The deep hash deduplication algorithm proposed in this paper can be applied to streamline the dataset for concrete visual inspection, which is of great significance for the efficient stitching of subsequent images, the accurate detection of diseases, and the three-dimensional (3D) reconstruction of bridge piers.

关键词： Concrete visual inspection Self-supervised deep hashing image deduplication masked image modeling Contrastive learning

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：