检索结果-内蒙古大学图书馆

A Novel Divide and Conquer Solution for Long-term Video Salient Object Detection

Machine Intelligence Research 2024年第4期21卷 684-703页

作者： Yun-Xiao Li Cheng-Li-Zhao Chen Shuai Li Ai-Min Hao Hong Qin State Key Laboratory of Virtual Reality Technology and Systems Beihang UniversityBeijing 100191China College of Computer Science and Technology China University of Petroleum(East China)Qingdao 266580China Department of Computer Science Stony Brook UniversityNew York 11794USA

Recently,a new research trend in our video salient object detection(VSOD)research community has focused on enhancing the detection results via model self-fine-tuning using sparsely mined high-quality keyframes from the given *** such a learning scheme is generally effective,it has a critical limitation,i.e.,the model learned on sparse frames only possesses weak generalization *** situation could become worse on“long”videos since they tend to have intensive scene ***,in such videos,the keyframe information from a longer time span is less relevant to the previous,which could also cause learning conflict and deteriorate the model ***,the learning scheme is usually incapable of handling complex pattern *** solve this problem,we propose a divide-and-conquer framework,which can convert a complex problem domain into multiple simple ***,we devise a novel background consistency analysis(BCA)which effectively divides the mined frames into disjoint *** for each group,we assign an individual deep model on it to capture its key attribute during the fine-tuning *** the testing phase,we design a model-matching strategy,which could dynamically select the best-matched model from those fine-tuned ones to handle the given testing *** experiments show that our method can adapt severe background appearance variation coupling with object movement and obtain robust saliency detection compared with the previous scheme and the state-of-the-art methods.

关键词： Video salient object detection background consistency analysis weakly supervised learning long-term information background shift.

来源：评论

学校读者我要写书评

暂无评论

An energy constraint position-based dynamics with corrected SPH kernel

引用

Science China(Information Sciences) 2023年第1期66卷 144-157页

作者： Wei CAO Luan LYU Zhixin YANG Enhua WU College of Computer Science and Technology China University of Petroleum Faculty of Science and Technology University of Macau State Key Laboratory of Internet of Things for Smart City University of Macau State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Guangzhou Greater Bay Area Virtual Reality Research Institute

We introduce an improved position-based dynamics method with corrected smoothed particle hydrodynamics(SPH) kernel to simulate deformable solids. Using a strain energy constraint that follows the continuum mechanics, the method can maintain the efficiency and stability of the position-based approach while improving the physical plausibility of the simulation. We can easily simulate the behavior of anisotropic and plastic materials because the method is based on physics. Unlike the previous position-based simulations of continuous materials, we use weakly structured particles to discretize the model for the convenience of deformable object cutting. In this case, a corrected SPH kernel function is adopted to measure the deformation gradient and calculate the strain energy on each particle. We also propose a solution for the interparticle inversion and penetration in large deformation. To perform complex interaction scenarios, we provide a simple method for collision detection. We demonstrate the flexibility, efficiency, and robustness of the proposed method by simulating various scenes, including anisotropic elastic deformation, plastic deformation, model cutting, and large-scale elastic collision.

关键词： position-based dynamics deformable solids continuum mechanics

来源：评论

学校读者我要写书评

暂无评论

Self-adjustable hyper-graphs for video pose estimation based on spatial-temporal subspace construction

引用

Science China(Information Sciences) 2022年第3期65卷 260-262页

作者： Jizhou MA Shuai LI Hong QIN Aimin HAO Qinping ZHAO State Key Laboratory of Virtual Reality Technology and Systems Beihang University Department of Computer Science Stony Brook University

Dear editor,In recent years many supervised video pose estimation methods have achieved growing successes based on well-labeled training datasets. Nonetheless, when facing roughly-labeled training data, it still remains challenging to intrinsically encode the video contents' spatial-temporal coherency for robust video pose *** researches aimed to directly improve and refine the existing confidence maps by combining the spatial-temporal structure models [1, 2]. Li et al.

关键词： Video Pose Estimation Self-adjustable Hyper-graph Spatial-temporal Subspace Exploration Maximum Matching Subspace Operator Action Pattern

来源：评论

学校读者我要写书评

暂无评论

Discriminative feature encoding for intrinsic image decomposition

引用

Computational Visual Media 2023年第3期9卷 597-618页

作者： Zongji Wang Yunfei Liu Feng Lu Key Laboratory of Network Information System Technology(NIST) Aerospace Information Research InstituteChinese Academy of SciencesBeijing 100190China State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijing 100191China Peng Cheng Laboratory Shenzhen 518000China

Intrinsic image decomposition is an important and long-standing computer vision *** an input image,recovering the physical scene properties is *** physically motivated priors have been used to restrict the solution space of the optimization problem for intrinsic image *** work takes advantage of deep learning,and shows that it can solve this challenging computer vision problem with high *** focus lies in the feature encoding phase to extract discriminative features for different intrinsic layers from an input *** achieve this goal,we explore the distinctive characteristics of different intrinsic components in the high-dimensional feature embedding *** define feature distribution divergence to efficiently separate the feature vectors of different intrinsic *** feature distributions are also constrained to fit the real ones through a feature distribution *** addition,a data refinement approach is provided to remove data inconsistency from the Sintel dataset,making it more suitable for intrinsic image *** method is also extended to intrinsic video decomposition based on pixel-wise correspondences between adjacent *** results indicate that our proposed network structure can outperform the existing state-of-the-art.

关键词： intrinsic image decomposition deep learning feature distribution data refinement

来源：评论

学校读者我要写书评

暂无评论

Uncertainty-Estimation With Normalized Logits For Out-of-distribution Detection

Uncertainty-Estimation With Normalized Logits For Out-of-dis...

引用

2023 International Conference on computer, Artificial Intelligence, and Control Engineering, CAICE 2023

作者： Huang, Mouxiao Qiao, Yu The Guangdong Provincial Key Laboratory of Computer Vision and Virtual Reality Technology Shenzhen Institute of Advanced Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

ISBN: (纸本)9781510665040

Out-of-distribution (OOD) detection is critical for preventing deep learning models from making incorrect predictions, especially in safety-critical applications such as medical diagnosis and autonomous driving. However, neural networks often suffer from overconfidence, making high confidence predictions for OOD data that are never seen during training and may be irrelevant to training data. Determining the reliability of the prediction is still a difficult and challenging task. To address this challenge, we propose a new method called Uncertainty-Estimation with Normalized Logits (UE-NL) for robust learning and OOD detection. The method has three main benefits: (1) Neural networks with UE-NL treat every in-distribution (ID) sample equally by predicting the uncertainty score of input data and the uncertainty is added into softmax function to adjust the learning strength of easy and hard samples during training phase, making the model learn robustly and accurately. (2) UE-NL enforces a constant vector norm on the logits to decouple the effect of the increasing output’s norm from optimization process, which causes the overconfidence issue to some extent. (3) UE-NL provides a new metric, the magnitude of uncertainty score, to detect OOD data. Experiments demonstrate that UE-NL outperforms existing methods on common OOD benchmarks and is more robust to noisy ID data that may be misjudged as OOD data by other methods. © 2023 SPIE.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Transforming vision Transformer: Towards Efficient Multi-Task Asynchronous Learning 38

Transforming Vision Transformer: Towards Efficient Multi-Tas...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zhong, Hanwen Chen, Jiaxin Zhang, Yutong Huang, Di Wang, Yunhong State Key Laboratory of Virtual Reality Technology and Systems Beihang University China School of Computer Science and Engineering Beihang University Beijing China

Multi-Task Learning (MTL) for vision Transformer aims at enhancing the model capability by tackling multiple tasks simultaneously. Most recent works have predominantly focused on designing Mixture-of-Experts (MoE) structures and integrating Low-Rank Adaptation (LoRA) to efficiently perform multi-task learning. However, their rigid combination hampers both the optimization of MoE and the effectiveness of reparameterization of LoRA, leading to sub-optimal performance and low inference speed. In this work, we propose a novel approach dubbed Efficient Multi-Task Learning (EMTAL) by transforming a pre-trained vision Transformer into an efficient multi-task learner during training, and reparameterizing the learned structure for efficient inference. Specifically, we firstly develop the MoEfied LoRA structure, which decomposes the pre-trained Transformer into a low-rank MoE structure and employ LoRA to fine-tune the parameters. Subsequently, we take into account the intrinsic asynchronous nature of multi-task learning and devise a learning Quality Retaining (QR) optimization mechanism, by leveraging the historical high-quality class logits to prevent a well-trained task from performance degradation. Finally, we design a router fading strategy to integrate the learned parameters into the original Transformer, archiving efficient inference. Extensive experiments on public benchmarks demonstrate the superiority of our method, compared to the state-of-the-art multi-task learning approaches. The project page is available at https://***/Yewen1486/EMTAL. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Subthreshold Depression Recognition and Correlation Study from Pulse Condition via Stacking Ensemble Algorithm 1st

Subthreshold Depression Recognition and Correlation Study f...

引用

1st International Conference on Extended reality, ICXR 2024

作者： Jiang, Han Li, Ming Gao, Yang Li, Peiru State Key Laboratory of Virtual Reality Technology and Systems Beihang University Beijing China School of Computer Science and Engineering Beihang University Beijing China

ISBN: (纸本)9789819636785

Subthreshold depression, a transition state to depression, seriously hinders the early diagnosis of depression. Current studies mostly use heterogeneous definitions of subthreshold depression, making the results of such meta-analyses questionable. Therefore, it is of vital significance to develop an objective method for the diagnosis of subthreshold depression based on objective criteria. In traditional Chinese medicine (TCM), symptoms similar to subthreshold depression have been extensively explored. However, diagnostic methods in TCM still depend heavily on the experience of doctors and lack integration with modern diagnostic techniques, which makes it challenging to explain the pathogenesis of subthreshold depression. Consequently, we propose an explainable framework, based on a stacking ensemble algorithm, for subthreshold depression recognition from biomarkers in the pulse waveform and concepts of pulse in TCM. In this method, Naive Bayes, Random Forest, Extremely Randomized Trees, Categorical Boosting and Logistic Regression are chosen as basic learners, and XGBoost is selected as the meta-classifier. Based on the five-fold cross-validation method, grid search method and repetition of training, the stacking ensemble model shows superiority on most performance evaluation metrics including AUC, F1 scores, MCC, precision and sensitivity. Besides, by analyzing the Adjusted Odds Ratio of features in the pulse waveform, we obtained four features that have a high correlation with the occurrence of subthreshold depression and derived physiological changes in patients with subthreshold depression based on their physiological significance. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Diagnosis

来源：评论

学校读者我要写书评

暂无评论

A survey for light field super-resolution

引用

High-Confidence Computing 2024年第1期4卷 118-129页

作者： Mingyuan Zhao Hao Sheng Da Yang Sizhe Wang Ruixuan Cong Zhenglong Cui Rongshan Chen Tun Wang Shuai Wang Yang Huang Jiahao Shen State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijing 100191China Key Laboratory of Data Science and Intelligent Computing International Innovation InstituteBeihang UniversityHangzhou 311115China State Key Laboratory of Software Development Environment School of Computer Science and EngineeringBeihang UniversityBeijing 100191China

Compared to 2D imaging data,the 4D light field(LF)data retains richer scene’s structure information,which can significantly improve the computer’s perception capability,including depth estimation,semantic segmentation,and LF ***,there is a contradiction between spatial and angular resolution during the LF image acquisition *** overcome the above problem,researchers have gradually focused on the light field super-resolution(LFSR).In the traditional solutions,researchers achieved the LFSR based on various optimization frameworks,such as Bayesian and Gaussian *** learning-based methods are more popular than conventional methods because they have better performance and more robust generalization *** this paper,the present approach can mainly divided into conventional methods and deep learning-based *** discuss these two branches in light field spatial super-resolution(LFSSR),light field angular super-resolution(LFASR),and light field spatial and angular super-resolution(LFSASR),***,this paper also introduces the primary public datasets and analyzes the performance of the prevalent approaches on these ***,we discuss the potential innovations of the LFSR to propose the progress of our research field.

关键词： Light field super-resolution Convolutional neural network Transformer Sub-aperture image Epipolar-plane image

来源：评论

学校读者我要写书评

暂无评论

Deblurring Neural Radiance Fields with Event-driven Bundle Adjustment 24

Deblurring Neural Radiance Fields with Event-driven Bundle A...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Qi, Yunshan Zhu, Lin Zhao, Yifan Bao, Nan Li, Jia State Key Laboratory of Virtual Reality Technology and Systems Scse Beihang University Beijing China School of Computer Science and Technology Beijing Institute of Technology Beijing China

ISBN: (纸本)9798400706868

Neural Radiance Fields (NeRF) achieves impressive 3D representation learning and novel view synthesis results with high-quality multi-view images as input. However, motion blur in images often occurs in low-light and high-speed motion scenes, which significantly degrades the reconstruction quality of NeRF. Previous deblurring NeRF methods struggle to estimate pose and lighting changes during the exposure time, making them unable to accurately model the motion blur. The bio-inspired event camera measuring intensity changes with high temporal resolution makes up this information deficiency. In this paper, we propose Event-driven Bundle Adjustment for Deblurring Neural Radiance Fields (EBAD-NeRF) to jointly optimize the learnable poses and NeRF parameters by leveraging the hybrid event-RGB data. An intensity-change-metric event loss and a photo-metric blur loss are introduced to strengthen the explicit modeling of camera motion blur. Experiments on both synthetic and real-captured data demonstrate that EBAD-NeRF can obtain accurate camera trajectory during the exposure time and learn a sharper 3D representations compared to prior works. © 2024 ACM.

关键词： 3D modeling

来源：评论

学校读者我要写书评

暂无评论

3D scene graph prediction from point clouds

引用

virtual reality & Intelligent Hardware 2022年第1期4卷 76-88页

作者： Fanfan WU Feihu YAN Weimin SHI Zhong ZHOU State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and EngineeringBeihang UniversityBeijing 100191China

Background In this study,we propose a novel 3D scene graph prediction approach for scene understanding from point *** It can automatically organize the entities of a scene in a graph,where objects are nodes and their relationships are modeled as *** specifically,we employ the DGCNN to capture the features of objects and their relationships in the scene.A Graph Attention Network(GAT)is introduced to exploit latent features obtained from the initial estimation to further refine the object arrangement in the graph structure.A one loss function modified from cross entropy with a variable weight is proposed to solve the multi-category problem in the prediction of object and *** Experiments reveal that the proposed approach performs favorably against the state-of-the-art methods in terms of predicate classification and relationship prediction and achieves comparable performance on object classification *** The 3D scene graph prediction approach can form an abstract description of the scene space from point clouds.

关键词： Scene understanding 3D scene graph Point cloud DGCNN GAT

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：