检索结果-内蒙古大学图书馆

27th International Conference on Pattern Recognition, ICPR 2024

作者： Hirner, Dominik Fraundorfer, Friedrich Institute of Computer Graphics and Vision Graz University of Technology Graz Austria Cologne Germany

ISBN: (纸本)9783031781919

Stereo estimation has made many advancements in recent years with the introduction of deep-learning. However the traditional supervised approach to deep-learning requires the creation of accurate and plentiful ground-truth data, which is expensive to create and not available in many situations. This is especially true for remote sensing applications, where there is an excess of available data without proper ground truth. To tackle this problem, we propose a self-supervised CNN with self-improving adaptive abilities. In the first iteration, the created disparity map is inaccurate and noisy. Leveraging the left-right consistency check, we get a sparse but more accurate disparity map which is used as an initial pseudo ground-truth. This pseudo ground-truth is then adapted and updated after every epoch in the training step of the network. We use the sum of inconsistent points in order to track the network convergence. The code for our method will be made available after acceptance at https://***/thedodo/SAda-Net. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Anytime Optimal Trajectory Repairing for Autonomous Vehicles

IEEE Open Journal of Intelligent Transportation Systems

引用

IEEE Open Journal of Intelligent Transportation Systems 2025年 6卷 537-553页

作者： Tong, Kailin Steinberger, Martin Horn, Martin Solmaz, Selim Watzenig, Daniel Virtual Vehicle Research GmbH Graz 8010 Austria Graz University of Technology Institute of Automation and Control Graz 8010 Austria Graz University of Technology Institute of Computer Graphics and Vision Graz 8010 Austria

Adapting to dynamically changing situations remains a pivotal challenge for automated driving systems, which demand robust and efficient solutions. Occasional perception errors inherent in artificial intelligence further complicate the task. Whereas traditional motion planning algorithms address this challenge by replanning the entire trajectory, a significantly more efficient strategy is to repair only the flawed segments. Our paper introduces a groundbreaking approach by formulating an optimal trajectory repairing problem and proposing an innovative and efficient framework for critical timing detection and trajectory repairing. This trajectory repairing specifically employs Bernstein basis polynomials in both 2D distance-time and 3D spatiotemporal spaces. A distinctive feature of our method is the use of an anytime grid search to determine a sub-optimal time-to-repair, which contrasts with previous methods that relied on manually tuned or fixed repair times, limiting both flexibility and robustness. A statistical analysis of 100 scenarios demonstrates that our trajectory-repairing framework outperforms the path-speed decoupled repairing framework in terms of scenario success rate. Furthermore, we introduce a novel algorithm for driving corridor generation that more accurately approximates the collision-free space than state-of-the-art work. The proposed approach has broad potential for application in embedded systems across various autonomous platforms. © 2020 IEEE.

关键词： Autonomous vehicles collision avoidance trajectory planning vehicle safety

来源：评论

学校读者我要写书评

暂无评论

Bigger Isn’t Always Better: Towards a General Prior for Medical Image Reconstruction

Bigger Isn’t Always Better: Towards a General Prior for M...

引用

46th Annual Conference of the German Association for Pattern Recognition, DAGM-GCPR 2024

作者： Glaszner, Lukas Zach, Martin Institute of Computer Graphics and Vision Graz University of Technology Inffeldgasse 16/II Graz 8010 Austria

ISBN: (纸本)9783031851803

Diffusion models have been successfully applied to many inverse problems, including MRI and CT reconstruction. Researchers typically re-purpose models originally designed for unconditional sampling without modifications. Using two different posterior sampling algorithms, we show empirically that such large networks are not necessary. Our smallest model, effectively a ResNet, performs almost as good as an attention U-Net on in-distribution reconstruction, while being significantly more robust towards distribution shifts. Furthermore, we introduce models trained on natural images and demonstrate that they can be used in both MRI and CT reconstruction, out-performing model trained on medical images in out-of-distribution cases. As a result of our findings, we strongly caution against simply re-using very large networks and encourage researchers to adapt the model complexity to the respective task. Moreover, we argue that a key step towards a general diffusion-based prior is training on natural images. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Data-driven prior Medical image reconstruction Score-based model

来源：评论

学校读者我要写书评

暂无评论

LA-CaRe-CNN: Cascading Refinement CNN for Left Atrial Scar Segmentation 1st

LA-CaRe-CNN: Cascading Refinement CNN for Left Atrial Scar ...

引用

1st MICCAI Challenge Comprehensive Analysis and Computing of Real-World Medical Images, CARE 2024 Held in Conjunction with 27th International Conference on Medical Image Computing and computer-Assisted Intervention, MICCAI 2024

作者： Thaler, Franz Štern, Darko Plank, Gernot Urschler, Martin Gottfried Schatz Research Center: Medical Physics and Biophysics Medical University of Graz Graz Austria Institute of Computer Graphics and Vision Graz University of Technology Graz Austria AVL List GmbH Graz Austria Institute for Medical Informatics Statistics and Documentation Medical University of Graz Graz Austria

ISBN: (纸本)9783031870088

Atrial fibrillation (AF) represents the most prevalent type of cardiac arrhythmia for which treatment may require patients to undergo ablation therapy. In this surgery cardiac tissues are locally scarred on purpose to prevent electrical signals from causing arrhythmia. Patient-specific cardiac digital twin models show great potential for personalized ablation therapy, however, they demand accurate semantic segmentation of healthy and scarred tissue typically obtained from late gadolinium enhanced (LGE) magnetic resonance (MR) scans. In this work we propose the Left Atrial Cascading Refinement CNN (LA-CaRe-CNN), which aims to accurately segment the left atrium as well as left atrial scar tissue from LGE MR scans. LA-CaRe-CNN is a 2-stage CNN cascade that is trained end-to-end in 3D, where Stage 1 generates a prediction for the left atrium, which is then refined in Stage 2 in conjunction with the original image information to obtain a prediction for the left atrial scar tissue. To account for domain shift towards domains unknown during training, we employ strong intensity and spatial augmentation to increase the diversity of the training dataset. Our proposed method based on a 5-fold ensemble achieves great segmentation results, namely, 89.21% DSC and 1.6969 mm ASSD for the left atrium, as well as 64.59% DSC and 91.80% G-DSC for the more challenging left atrial scar tissue. Thus, segmentations obtained through LA-CaRe-CNN show great potential for the generation of patient-specific cardiac digital twin models and downstream tasks like personalized targeted ablation therapy to treat AF. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Ablation

来源：评论

学校读者我要写书评

暂无评论

Augmentation-Based Domain Generalization and Joint Training from Multiple Source Domains for Whole Heart Segmentation 1st

Augmentation-Based Domain Generalization and Joint Training...

引用

ISBN: (纸本)9783031870088

As the leading cause of death worldwide, cardiovascular diseases motivate the development of more sophisticated methods to analyze the heart and its substructures from medical images like Computed Tomography (CT) and Magnetic Resonance (MR). Semantic segmentations of important cardiac structures that represent the whole heart are useful to assess patient-specific cardiac morphology and pathology. Furthermore, accurate semantic segmentations can be used to generate cardiac digital twin models which allows e.g. electrophysiological simulation and personalized therapy planning. Even though deep learning-based methods for medical image segmentation achieved great advancements over the last decade, retaining good performance under domain shift - i.e. when training and test data are sampled from different data distributions - remains challenging. In order to perform well on domains known at training-time, we employ a (1) balanced joint training approach that utilizes CT and MR data in equal amounts from different source domains. Further, aiming to alleviate domain shift towards domains only encountered at test-time, we rely on (2) strong intensity and spatial augmentation techniques to greatly diversify the available training data. Our proposed whole heart segmentation method, a 5-fold ensemble with our contributions, achieves the best performance for MR data overall and a performance similar to the best performance for CT data when compared to a model trained solely on CT. With 93.33% DSC and 0.8388 mm ASSD for CT and 89.30% DSC and 1.2411 mm ASSD for MR data, our method demonstrates great potential to efficiently obtain accurate semantic segmentations from which patient-specific cardiac twin models can be generated. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Bigger Isn’t Always Better: Towards a General Prior for Medical Image Reconstruction

arXiv

引用

arXiv 2025年

作者： Glaszner, Lukas Zach, Martin Institute of Computer Graphics and Vision Graz University of Technology Inffeldgasse 16/II Graz8010 Austria Biomedical Imaging Group École polytechnique fédérale de Lausanne Lausanne1015 Switzerland Center for Biomedical Imaging Lausanne1015 Switzerland

Diffusion model have been successfully applied to many inverse problems, including MRI and CT reconstruction. Researchers typically re-purpose models originally designed for unconditional sampling without modifications. Using two different posterior sampling algorithms, we show empirically that such large networks are not necessary. Our smallest model, effectively a ResNet, performs almost as good as an attention U-Net on in-distribution reconstruction, while being significantly more robust towards distribution shifts. Furthermore, we introduce models trained on natural images and demonstrate that they can be used in both MRI and CT reconstruction, out-performing model trained on medical images in out-of-distribution cases. As a result of our findings, we strongly caution against simply re-using very large networks and encourage researchers to adapt the model complexity to the respective task. Moreover, we argue that a key step towards a general diffusion-based prior is training on natural images. © 2025, CC BY-SA.

关键词： Medical imaging

来源：评论

学校读者我要写书评

暂无评论

Gaussian Process Emulators for Few-Shot Segmentation in Cardiac MRI

Gaussian Process Emulators for Few-Shot Segmentation in Ca...

引用

15th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2024, Held in Conjunction with MICCAI 2024

作者： Viti, Bruno Thaler, Franz Kapper, Kathrin Lisa Urschler, Martin Holler, Martin Karabelas, Elias Department of Mathematics and Scientific Computing University of Graz Graz Austria Institute for Medical Informatics Statistics and Documentation Medical University of Graz Graz Austria Institute of Computer Graphics and Vision Graz University of Technology Graz Austria Gottfried Schatz Research Center: Medical Physics and Biophysics Medical University of Graz Graz Austria BioTechMed-Graz Graz Austria

ISBN: (纸本)9783031877551

Segmentation of cardiac magnetic resonance images (MRI) is crucial for the analysis and assessment of cardiac function, helping to diagnose and treat various cardiovascular diseases. Most recent techniques rely on deep learning and usually require an extensive amount of labeled data. To overcome this problem, few-shot learning has the capability of reducing data dependency on labeled data. In this work, we introduce a new method that merges few-shot learning with a U-Net architecture and Gaussian Process Emulators (GPEs), enhancing data integration from a support set for improved performance. GPEs are trained to learn the relation between the support images and the corresponding masks in latent space, facilitating the segmentation of unseen query images given only a small labeled support set at inference. We test our model with the M&Ms-2 public dataset to assess its ability to segment the heart in cardiac magnetic resonance imaging from different orientations, and compare it with state-of-the-art unsupervised and few-shot methods. Our architecture shows higher DICE coefficients compared to these methods, especially in the more challenging setups where the size of the support set is considerably small. The code is available on GitLab (https://***/bruno_viti/gpe_4_cardiac_fss). © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Gaussian distribution

来源：评论

学校读者我要写书评

暂无评论

Intraoperative measurements in stapedotomy using 3D stereo imaging for optimal prosthesis length selection

引用

computers in Biology and Medicine 2025年 191卷 110233-110233页

作者： Wisotzky, Eric L. Lausch, Felix Kienle, Linus L. van Bonn-Ytrehus, Sara Hilsmann, Anna Eisert, Peter Schraven, Sebastian P. Mlynski, Robert Computer Vision & Graphics Vision and Imaging Technologies Fraunhofer Heinrich Hertz Institut HHI Berlin Germany Institute of Informatics Humboldt Universität zu Berlin Germany Klinik und Poliklinik für Hals-Nasen-Ohrenheilkunde Kopf- und Halschirurgie "Otto Körner" Universitätsmedizin Rostock Germany Department of Othorinolaryngology Head and Neck Surgery Interdisciplinary Skull Base Center Klinikum Fulda GAG University Medicine Marburg Campus Fulda Germany

The application of 3D stereoscopic imaging in surgery represents an innovative approach for anatomical measurements without the use of ionizing radiation or tools contacting the anatomy. Achieving precise intraoperative measurements is crucial in microsurgery, yet conventional methods often lack accuracy due to technical limitations in microscopic zoom lens systems. This study investigates 3D imaging within a digital microscope for stapedotomy, focusing on its accuracy and clinical applicability in selecting optimal prosthesis lengths. We present an optimized calibration scheme for stereoscopic zoom-focus systems across all focus settings, particularly at high magnification levels. A cohort of 23 patients underwent stapedotomy with stereo imaging for landmark annotation and prosthesis measurement. Our calibration method ensured sub-millimeter accuracy, achieving an average deviation of 0.2097 ± 0.1598 mm. The findings demonstrated a significant correlation between insertion depth and postoperative audiological outcomes. Audiological evaluations revealed a mean air-bone gap improvement of 19.1 dB [HL], validating the method's clinical efficacy. This technique offers a radiation-free, efficient alternative to conventional methods, integrating seamlessly into surgical workflows. The study highlights the potential of this imaging modality to enhance surgical precision and patient outcomes, setting the stage for future advancements in automated real-time measurements and broader clinical validation. © 2025 The Author(s)

关键词： Stereo image processing

来源：评论

学校读者我要写书评

暂无评论

Removing fences from sweep motion videos using global 3D reconstruction and fence-aware light field rendering

引用

Computational Visual Media 2019年第1期5卷 21-32页

作者： Chanya Lueangwattana Shohei Mori Hideo Saito Department of Science and Technology Keio University Institute of Computer Graphics and Vision Graz University of Technology

Diminishing the appearance of a fence in an image is a challenging research area due to the characteristics of fences(thinness, lack of texture, etc.) and the need for occluded background restoration. In this paper, we describe a fence removal method for an image sequence captured by a user making a sweep motion, in which occluded background is potentially observed. To make use of geometric and appearance information such as consecutive images, we use two well-known approaches: structure from motion and light field rendering. Results using real image sequences show that our method can stably segment fences and preserve background details for various fence and background combinations. A new video without the fence, with frame coherence, can be successfully provided.

关键词： video fence video repair diminished reality(DR) structure from motion(SfM) light field rendering(LFR)

来源：评论

学校读者我要写书评

暂无评论

Motion Estimation with Non-Local Total Variation Regularization

Motion Estimation with Non-Local Total Variation Regularizat...

引用

23rd IEEE Conference on computer vision and Pattern Recognition (CVPR)

作者： Werlberger, Manuel Pock, Thomas Bischof, Horst Institute for Computer Graphics and Vision Austria

ISBN: (纸本)9781424469840

State-of-the-art motion estimation algorithms suffer from three major problems: Poorly textured regions, occlusions and small scale image structures. Based on the Gestalt principles of grouping we propose to incorporate a low level image segmentation process in order to tackle these problems. Our new motion estimation algorithm is based on non-local total variation regularization which allows us to integrate the low level image segmentation process in a unified variational framework. Numerical results on the Middlebury optical flow benchmark data set demonstrate that we can cope with the aforementioned problems.

关键词： Motion estimation

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：