检索结果-内蒙古大学图书馆

SurgicalGaussian: Deformable 3D Gaussians for High-Fidelity Surgical Scene Reconstruction

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xie, Weixing Yao, Junfeng Cao, Xianpeng Lin, Qiqin Tang, Zerui Dong, Xiao Guo, Xiaohu Center for Digital Media Computing School of Film School of Informatics Xiamen University Xiamen China National Institute for Data Science in Health and Medicine Xiamen University China Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage of Fujian and Taiwan Ministry of Culture and Tourism China Guangdong Provincial Key Laboratory IRADS Department of Computer Science BNU-HKBU United International College Zhuhai China Department of Computer Science The University of Texas at Dallas Dallas United States

Dynamic reconstruction of deformable tissues in endoscopic video is a key technology for robot-assisted surgery. Recent reconstruction methods based on neural radiance fields (NeRFs) have achieved remarkable results in the reconstruction of surgical scenes. However, based on implicit representation, NeRFs struggle to capture the intricate details of objects in the scene and cannot achieve real-time rendering. In addition, restricted single view perception and occluded instruments also propose special challenges in surgical scene reconstruction. To address these issues, we develop SurgicalGaussian, a deformable 3D Gaussian Splatting method to model dynamic surgical scenes. Our approach models the spatio-temporal features of soft tissues at each time stamp via a forward-mapping deformation MLP and regularization to constrain local 3D Gaussians to comply with consistent movement. With the depth initialization strategy and tool mask-guided training, our method can remove surgical instruments and reconstruct high-fidelity surgical scenes. Through experiments on various surgical videos, our network outperforms existing method on many aspects, including rendering quality, rendering speed and GPU usage. The project page can be found at https://***. © 2024, CC BY-SA.

关键词： Tissue

Context-aware Multimodal AI Reveals Hidden Pathways in Five Centuries of Art Evolution

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Kim, Jin Lee, Byunghwee You, Taekho Yun, Jinhyuk Department of Intelligent Semiconductors Soongsil University Seoul06978 Korea Republic of School of Digital Humanities & Computational Social Sciences Korea Advanced Institute of Science and Technology Daejeon34141 Korea Republic of Luddy School of Informatics Computing and Engineering Indiana University BloomingtonIN47408 United States Institute for Social Data Science Pohang University of Science and Technology Pohang37673 Korea Republic of Center for Digital Humanities & Computational Social Sciences Korea Advanced Institute of Science and Technology Daejeon34141 Korea Republic of School of AI Convergence Soongsil University Seoul06978 Korea Republic of

The rise of multimodal generative AI is transforming the intersection of technology and art, offering deeper insights into large-scale artwork. Although its creative capabilities have been widely explored, its potential to represent artwork in latent spaces remains underexamined. We use cutting-edge generative AI, specifically Stable Diffusion, to analyze 500 years of Western paintings by extracting two types of latent information with the model: formal aspects (e.g., colors) and contextual aspects (e.g., subject). Our findings reveal that contextual information differentiates between artistic periods, styles, and individual artists more successfully than formal elements. Additionally, using contextual keywords extracted from paintings, we show how artistic expression evolves alongside societal changes. Our generative experiment, infusing prospective contexts into historical artworks, successfully reproduces the evolutionary trajectory of artworks, highlighting the significance of mutual interaction between society and art. This study demonstrates how multimodal AI expands traditional formal analysis by integrating temporal, cultural, and historical contexts. Copyright © 2025, The Authors. All rights reserved.

关键词： Generative adversarial networks

A novel automatic wind power prediction framework based on multi-time scale and temporal attention mechanisms

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Jiang, Meiyu Shen, Jun Jiang, Xuetao Luo, Lihui Zhou, Rui Zhou, Qingguo School of Information Science and Engineering Lanzhou University Gansu Lanzhou730000 China School of Computing and Information Technology University of Wollongong WollongongNSW2522 Australia Gansu Provincial Industry Technology Center of Intelligent Equipment & Big Data for Disaster Prevention Northwest Institute of Eco-Environment and Resources Chinese Academy of Sciences Gansu Lanzhou730000 China

Wind energy is a widely distributed, renewable, and environmentally friendly energy source that plays a crucial role in mitigating global warming and addressing energy shortages. Nevertheless, wind power generation is characterized by volatility, intermittence, and randomness, which hinder its ability to serve as a reliable power source for the grid. Accurate wind power forecasting is crucial for developing a new power system that heavily relies on renewable energy sources. However, traditional wind power forecasting systems primarily focus on ultra-short-term or short-term forecasts, limiting their ability to address the diverse adjustment requirements of the power system simultaneously. To overcome these challenges, We propose an automatic framework capable of forecasting wind power across multi-time scale. The framework based on the tree-structured Parzen estimator (TPE) and temporal fusion transformer (TFT) that can provide ultra-short-term, short-term and medium-term wind power forecasting power. Our approach employs the TFT for wind power forecasting and categorizes features based on their properties. Additionally, we introduce a generic algorithm to simultaneously fine-tune the hyperparameters of the decomposition method and model. We evaluate the performance of our framework by conducting ablation experiments using three commonly used decomposition algorithms and six state-of-the-art models for forecasting multi-time scale. The experimental results demonstrate that our proposed method considerably improves prediction accuracy on the public dataset Engie https://***. Compared to the second-best state-of-the-art model, our approach exhibits a reduction of 31.75% and 28.74% in normalized mean absolute error (nMAE) for 24-hour forecasting, and 20.79% and 16.93% in nMAE for 48-hour forecasting, respectively. © 2023, CC BY.

关键词： Wind power

PD-SDF: Dynamic Surface Reconstruction Based on Plane Decomposition for Single View RGB-D Videos

学校读者我要写书评

暂无评论

PD-SDF: Dynamic Surface Reconstruction Based on Plane Decomp...

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Jinwen Li Weixing Xie Junfeng Yao Shaoqi Wu Youhong Peng Mengyuan Ge Xiao Dong Center for Digital Media Computing School of Film School of Informatics Xiamen University National Institute for Data Science in Health and Medicine Xiamen University Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage of Fujian and Taiwan Ministry of Culture and Tourism Jiujiang Research Institute of Xiamen University Jiujiang China Guangdong Provincial/Zhuhai Key Laboratory of IRADS BNU-HKBU United International College Zhuhai China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Surface reconstruction of dynamic scenes from single view videos is a challenging task due to the highly ill-posed and under-constrained nature. Existing single view reconstruction methods suffer from severe quality issues, such as surface distortion and mesh adherison. In this paper, we propose an efficient dynamic representation network, PD-SDF, which consists of a 4D motion field and a 3D geometry field. The explicit disentanglement of motion and geometry based on planar factorization guarantees the mesh consistency across frames. Specifically, to address the mesh adhesion problem, we design a depth-guided sampling strategy to focus on optimizing the SDF field near the object surface. Due to insufficient geometric cues, we design various regularization strategies to constrain smoothness and topological correcness of the scene geometry. Extensive experiments show that our method outperforms existing methods in both appearance and geometry reconstruction. The project page: https://***/

关键词： Geometry Training Surface reconstruction Three-dimensional displays Noise Reconstruction algorithms Stability analysis Speech processing Surface treatment Videos

Optimal Quantum State Tomography via Weak Value

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhu, Xuanmin Zhang, Dezheng Gao, Runping Wei, Qun Liu, Lixia Luo, Zijiang Center for Quantum Information School of Information Guizhou University of Finance and Economics Guizhou Guiyang550025 China International Joint Research Center for Data Science and High-Performance Computing School of Information Guizhou University of Finance and Economics Guizhou Guiyang550025 China School of Physics and Optoelectronic Engineering Xidian University Xi’an710071 China School of Mathematics and Statistics Xidian University Xi’an710071 China Institute of Intelligent Manufacturing Shunde Polytechnic Guangdong Shunde 528300 China

To improve the efficiency of the state tomography strategy via weak value, we have searched the optimal coupling strength between the system and measuring device. For an arbitrary d-dimensional quantum system, the optimal strengths being used in measuring the real and imaginary parts of the density matrix are obtained. The optimal efficiency of the state tomography has also been studied by using mean square error. The minimal mean square errors in the reconstructed density matrices have been derived. The state tomography strategy studied in this article may be useful in the measurement of the unknown quantum states. Copyright © 2024, The Authors. All rights reserved.

关键词： Mean square error

A Multi-Stage Deep Learning Pipeline for Crack Detection and Segmentation in High-Speed Rail Brake Discs

学校读者我要写书评

暂无评论

A Multi-Stage Deep Learning Pipeline for Crack Detection and...

Consumer Technology (ICCT-Pacific), International Conference on

作者： Ching-Heng Cheng Yu-Fan Lin Chia-Ming Lee Chih-Chung Hsu Yu-Lin Shih Ting Chin Department of Statistics Institute of Data Science National Cheng Kung Univ. Tainan Taiwan Department of Intelligent Technology Systems Institute of Miin Wu School of Computing National Cheng Kung Univ Tainan Taiwan Institute of Intelligent Systems College of Artificial Intelligence National Yang Ming Chiao Tung Univ. Hsinchu Taiwan Innovation Logistics System Technology Dept. Smart Logistics & Supply Chain Integration Service Division Service Systems Technology Center Industrial Technology Research Institute Hsinchu Taiwan

ISBN: (数字)9798331504120

ISBN: (纸本)9798331504137

The braking system of high-speed rail (HSR) is critical for passenger safety, but prolonged use of brake discs can lead to cracks, posing significant risks. Traditional manual inspection methods are time-consuming and labor-intensive. To address these challenges, we propose a multi-stage crack detection and segmentation system, a cascade mode to progressively identify cracks, leveraging deep learning techniques. The system includes automated circle detection, region cropping, and crack detection and segmentation to accurately identify cracks. Using a real-world dataset collected from HSR brake discs, our approach significantly improves data preprocessing efficiency and model performance. Furthermore, we evaluate the performance on un-seen data and compare our crack detection framework with other deep learning models. The experimental results demonstrate the system's effectiveness which is superior to other deep learning models, highlighting its potential for real-world deployment in HSR maintenance.

关键词： Deep learning Image segmentation Motion segmentation Computational modeling Pipelines Rail transportation data models Maintenance Safety Brakes

A Causal Adversarial Graph Neural Network for Multi-center Autism Spectrum Disorder Identification

学校读者我要写书评

暂无评论

SSRN

SSRN 2025年

作者： Zhang, Zhuan Wang, Lihui Chen, Qijian Wang, Li Jian, Caiqing Zhu, Yue-Min Wei, Hongjiang Engineering Research Center of Text Computing & Cognitive Intelligence Ministry of Education Key Laboratory of Intelligent Medical Image Analysis and Precise Diagnosis of Guizhou Province State Key Laboratory of Public Big Data College of Computer Science and Technology Guizhou University Guiyang550025 China University Lyon INSA Lyon CNRS Inserm IRP Metislab CREATIS UMR5220 U1206 Lyon69621 France

Due to the heterogeneous distributions of multi-center rs-fMRI data, currently it is still challenging to identify accurately autism spectrum disorder (ASD) patients from these heterogenous data. To deal with the influence of heterogeneous distributions on multi-center ASD diagnosis, a novel causal adversarial graph neural network (CAG) for multi-center ASD prediction is proposed in this work. CAG mainly consists of a graph filtering (GF) module and a learnable causal graph predictor (CGP). Specifically, the brain connection graph is first constructed from rs-fMRI data, after passing a GF module, the filtered graph is further divided by a causal subgraph and causally irrelevant subgraph. The ground-truth environmental labels and ASD labels are predicted from causal subgraph by a clustering method and a classification block, respectively. In addition, from the filtered graph and causally irrelevant subgraph, the environmental labels are predicted, while from the causally irrelevant subgraph the ASD labels are predicted. Assuming that the predicted environmental labels are independent to causal subgraph, while the predicted ASD labels are independent to causally irrelevant subgraph, accordingly, several adversarial losses are implemented by reversing the gradient. Through this adversarial learning strategy, the derived causal graph can predict multi-center ASD. The comparison experiments on both ABIDE-I and ABIDE-II datasets demonstrate that, the ACC and AUC of CAG has been improved by 3.78\% and 7.38\% at least comparing against the existing methods. Moreover, some abnormal functional connectivities in ASD group are identified from the causal subgraph which may provide useful imaging biomarkers for the early diagnosis of ASD. © 2025, The Authors. All rights reserved.

关键词： Graph neural networks

SoLar: sinkhorn label refinery for imbalanced partial-label learning 22

学校读者我要写书评

暂无评论

SoLar: sinkhorn label refinery for imbalanced partial-label ...

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Haobo Wang Mingxuan Xia Yixuan Li Yuren Mao Lei Feng Gang Chen Junbo Zhao Key Lab of Intelligent Computing based Big Data of Zhejiang Province Zhejiang University School of Software Technology Zhejiang University Department of Computer Sciences University of Wisconsin-Madison College of Computer Science Chongqing University and Center for Advanced Intelligence Project RIKEN

ISBN: (纸本)9781713871088

Partial-label learning (PLL) is a peculiar weakly-supervised learning task where the training samples are generally associated with a set of candidate labels instead of single ground truth. While a variety of label disambiguation methods have been proposed in this domain, they normally assume a class-balanced scenario that may not hold in many real-world applications. Empirically, we observe degenerated performance of the prior methods when facing the combinatorial challenge from the long-tailed distribution and partial-labeling. In this work, we first identify the major reasons that the prior work failed. We subsequently propose SoLar, a novel Optimal Transport-based framework that allows to refine the disambiguated labels towards matching the marginal class prior distribution. SoLar additionally incorporates a new and systematic mechanism for estimating the long-tailed class prior distribution under the PLL setup. Through extensive experiments, SoLar exhibits substantially superior results on standardized benchmarks compared to the previous state-of-the-art PLL methods.

关键词：

Photovoltaic Power Forecasting with Weather Conditioned Attention Mechanism

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Jiang, Xuetao Jiang, Meiyu Luo, Lihui Shen, Jun Zhou, Qingguo Zhou, Rui School of Information Science and Engineering Lanzhou University Tianshui South Road No.222 Gansu Lanzhou730000 China Gansu Provincial Industry Technology Center of Intelligent Equipment & Big Data for Disaster Prevention Northwest Institute of Eco-Environment and Resources Chinese Academy of Science Gansu Lanzhou730000 China School of Computing and Information Technology University of Wollongong WollongongNSW2522 Australia

Accurate photovoltaic (PV) generation forecasts can reduce power redeploy from the grid, thus increasing the supplier's profit in the day-ahead electricity market. However, the PV process is affected differently by various factors under different weather conditions, resulting in significantly different energy output curves. In this context, this paper proposes a day-ahead PV power forecasting method with weather conditioned attention mechanism. We propose a multi-stream attention fusion network (MSAFN) which utilizes an algorithm to derive the optimal decomposition algorithm for different weather conditions. The proposed conditional decomposition (CD) algorithm searches for the decomposition algorithms and corresponding hyperparameters of the prediction model, aiming to achieve the optimal prediction performance. The MSAFN incorporates multiple attention modules to learn the energy output patterns under various weather conditions. Notably, the attention modules adeptly learn patterns under diverse conditions, while simultaneously, the sharing of weights among the remaining components of the model effectively enhances prediction accuracy and facilitates a reduction in training time. We compare the state-of-the-art decomposition algorithms (VMD, EEMD and MSTL, etc.) and prediction models (BPN, LSTM, XGBoost and transformer, etc.) commonly used in PV prediction. The results show that the MSAFN model is more accurate than the models above, which has a noticeable improvement compared to other recent day-ahead PV predictions on DKASC. © 2023, The Authors. All rights reserved.

关键词： Long short-term memory