检索结果-南通市图书馆

Smoothness preserving layout for dynamic labels by hybrid optimization

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computational Visual Media 2022年第1期8卷 149-163页

作者： Yu He Guo-Dong Zhao song-hai zhang Department of Computer Science and Technology Tsinghua UniversityBeijingChina Department of Computer Science and Technology Tianjin University of TechnologyTianjinChina

Stable label movement and smooth label trajectory are critical for effective information *** label changes cannot be avoided by whatever forced directed methods due to the unreliability of resultant force or global optimization methods due to the complex trade-off on the different *** solve this problem,we proposed a hybrid optimization method by taking advantages of the merits of both *** first detect the spatial-temporal intersection regions from whole trajectories of the features,and initialize the layout by optimization in decreasing order by the number of the involved *** label movements between the spatial-temporal intersection regions are determined by force directed *** cope with some features with high speed relative to neighbors,we introduced a force from future,called temporal force,so that the labels of related features can elude ahead of time and retain smooth *** also proposed a strategy by optimizing the label layout to predict the trajectories of features so that such global optimization method can be applied to streaming data.

关键词： label layout smoothness preserving dynamic label forced based static optimization hybrid optimization

Deep image synthesis from intuitive user input:A review and perspectives

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Computational Visual Media 2022年第1期8卷 3-31页

作者： Yuan Xue Yuan-Chen Guo Han zhang Tao Xu song-hai zhang Xiaolei Huang College of Information Sciences and Technology the Pennsylvania State UniversityUniversity ParkPAUSA Department of Computer Science and Technology Tsinghua UniversityBeijingChinaand Beijing National Research Center for Information Science and Technology(BNRist)Tsinghua UniversityBeijingChina Google Brain Mountain ViewCAUSA Facebook Menlo ParkCAUSA.

In many applications of computer graphics,art,and design,it is desirable for a user to provide intuitive non-image input,such as text,sketch,stroke,graph,or layout,and have a computer system automatically generate photo-realistic images according to that *** classically,works that allow such automatic image content generation have followed a framework of image retrieval and composition,recent advances in deep generative models such as generative adversarial networks(GANs),variational autoencoders(VAEs),and flow-based methods have enabled more powerful and versatile image generation *** paper reviews recent works for image synthesis given intuitive user input,covering advances in input versatility,image generation methodology,benchmark datasets,and evaluation *** motivates new perspectives on input representation and interactivity,cross fertilization between major image generation paradigms,and evaluation and comparison of generation methods.

关键词： image synthesis intuitive user input deep generative models synthesized image quality evaluation

Focusing on your subject:Deep subject-aware image composition recommendation networks

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Computational Visual Media 2023年第1期9卷 87-107页

作者： Guo-Ye Yang Wen-Yang Zhou Yun Cai song-hai zhang Fang-Lue zhang BNRist Department of Computer Science and TechnologyTsinghua UniversityBeiing 100084China School of Engineering and Computer Science Victoria University of WellingtonWellington 6012New Zealand

Photo composition is one of the most important factors in the aesthetics of *** a popular application,composition recommendation for a photo focusing on a specific subject has been ignored by recent deep-learning-based composition recommendation *** this paper,we propose a subject-aware image composition recommendation method,SAC-Net,which takes an RGB image and a binary subject window mask as input,and returns good compositions as crops containing the *** model first determines candidate scores for all possible coarse cropping *** crops with high candidate scores are selected and further refined by regressing their corner points to generate the output recommended cropping *** final scores of the refined crops are predicted by a final score regression *** existing methods that need to preset several cropping windows,our network is able to automatically regress cropping windows with arbitrary aspect ratios and *** propose novel stability losses for maximizing smoothness when changing cropping windows along with view *** results show that our method outperforms state-of-the-art methods not only on the subject-aware image composition recommendation task,but also for general purpose composition *** also have designed a multistage labeling scheme so that a large amount of ranked pairs can be produced *** use this scheme to propose the first subject-aware composition dataset SACD,which contains 2777 images,and more than 5 million composition ranked *** SACD dataset is publicly available at https://***/SACD/.

关键词： subject-aware image composition image cropping deep learning recommendation

Learning Local Contrast for Crisp Edge Detection

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2023年第3期38卷 554-566页

作者：方晓楠张松海 Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

In recent years, the accuracy of edge detection on several benchmarks has been significantly improved by deep learning based methods. However, the prediction of deep neural networks is usually blurry and needs further post-processing including non-maximum suppression and morphological thinning. In this paper, we demonstrate that the blurry effect arises from the binary cross-entropy loss, and crisp edges could be obtained directly from deep convolutional neural networks. We propose to learn edge maps as the representation of local contrast with a novel local contrast loss. The local contrast is optimized in a stochastic way to focus on specific edge directions. Experiments show that the edge detection network trained with local contrast loss achieves a high accuracy comparable to previous methods and dramatically improves the crispness. We also present several applications of the crisp edges, including image completion, image retrieval, sketch generation, and video stylization.

关键词： edge detection image processing neural network local contrast

Local Homography Estimation on User-Specified Textureless Regions

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2022年第3期37卷 615-625页

作者：陈铮方晓楠张松海 Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

This paper presents a novel deep neural network for designated point tracking(DPT)in a monocular RGB video,*** concretely,the aim is to track four designated points correlated by a local homography on a textureless planar region in the *** can be applied to augmented reality and video editing,especially in the field of video *** methods predict the location of four designated points without appropriately considering the point *** solve this problem,VideoInNet predicts the motion of the four designated points correlated by a local homography within the heatmap prediction *** network refines the heatmaps of designated points through two *** the first stage,we introduce a context-aware and location-aware structure to learn a local homography for the designated plane in a supervised *** the second stage,we introduce an iterative heatmap refinement module to improve the tracking *** propose a dataset focusing on textureless planar regions,named ScanDPT,for training and *** show that the error rate of VideoInNet is about 29%lower than that of the state-of-the-art approach when testing in the first 120 frames of testing videos on ScanDPT.

关键词： homography estimation neural network designated point tracking(DPT)

Preparation and properties of ZrO_(2)-strengthened porous mullite insulation materials using Y_(2)O_(3) additive

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Iron and Steel Research(International) 2024年第6期31卷 1514-1521页

作者： Wang-zhi Yu Yi-hong Shi Ping Liu Yong-wan Wu Cheng-hao song Xian-gong Deng Xiang Ding song-lin Ran Yi zhang hai-jun zhang hai-liang Deng Advanced Ceramics Research Center School of Materials Science and EngineeringAnhui University of TechnologyMa'anshan 243002AnhuiChina The State Key Laboratory of Refractories and Metallurgy Wuhan University of Science and TechnologyWuhan 430081HubeiChina

ZrO_(2)-strengthened porous mullite insulation materials were prepared by foaming technology utilizing ZrSiO_(4) and Al_(2)O_(3) as primary materials and Y_(2)O_(3) as an *** effects of Y_(2)O_(3) contents on the phase composition,microstructure,mechanical properties,and heat conductivity of the porous mullite insulation materials were investigated.A suitable Y_(2)O_(3) content could promote phase transition of monoclinic ZrO_(2)(m-ZrO_(2))to tetragonal ZrO_(2)(t-ZrO_(2)),reduce pore size,and improve the strengths of as-prepared *** cold crushing strength and bending strength of as-prepared specimens with a 119µm spherical pore size using 6 wt.%Y_(2)O_(3) were 35.2 and 13.0 MPa,respectively,with a heat conductivity being only 0.248 W/(m K).

关键词： Porous mullite insulation material Zirconium oxide Y_(2)O_(3) Mechanical property Heat conductivity

Precise immobilization of metal single atoms into a porphyrinic metal-organic framework for an efficient alkene hydrosilylation

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Nano Research 2024年第7期17卷 5914-5921页

作者： Chun-Ying Chen Qi-Jie Mo Fu-Zhen Li hai-Li song Li zhang MOE Laboratory of Bioinorganic and Synthetic Chemistry Lehn Institute of Functional MaterialsSchool of ChemistrySun Yat-Sen UniversityGuangzhou 510006China

Alkene hydrosilylation is one of the most concise and atom-economical methods to synthesize organosilicon ***,we reported the precise immobilization of metal single atoms(M-SAs;M=Ru,Rh,Ir,Pd,Pt,and Au)into a porphyrinic metalorganic framework(MOF)of PCN-222(PCN=porous coordination network),and then applied the resultant MOF composites of M-SAs@PCN-222 to alkene *** solvent-free conditions,Pt-SAs@PCN-222 displayed an especially high catalytic efficiency with the turnover frequency up to 119 s−1 and the maximum turnover number of 906,250 at room *** and theoretical studies revealed that there existed strong interactions between Pt-SAs@PCN-222 and the substrates,which helped to condense the substrates in the cavities of the porous *** density functional theory calculations and molecular dynamics simulations disclosed that PCN-222 could transfer electrons to Pt-SAs to enhance the silane oxidative addition and drive the reaction to proceed smoothly via Chalk–Harrod pathway.

关键词： alkene hydrosilylation single atom catalysts metal-organic frameworks precision chemistry synergistic catalysis

Visualization of COVID-19 spread based on spread and extinction indexes

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2020年第6期63卷 220-222页

作者： song-hai zhang Yun CAI Jian LI Department of Computer Science and Technology Tsinghua University Beijing National Research Center for Information Science and Technology(BNRist)

From December 2019, Coronavirus disease 2019(COVID-19) rapidly spread from the city of Wuhan (Hubei Province) to all of China, which was exacerbated by mass population movements associated with the Chinese Spring Festival. To date, more than 80000 people have been infected,and more than 3000 have died. In addition,COVID-19 has already spread to 142 countries,

关键词： Visualization of COVID-19 spread based on spread and extinction indexes

Multi-Task Learning for Food Identification and Analysis with Deep Convolutional Neural Networks

同方期刊数据库评论

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2016年第3期31卷 489-500页

作者： Xi-Jin zhang Yi-Fan Lu song-hai zhang Department of Computer Science and Technology Tsinghua University Beijing 100084 China

In this paper, we proposed a multi-task system that can identify dish types, food ingredients, and cooking methods from food images with deep convolutional neural networks. We built up a dataset of 360 classes of different foods with at least 500 images for each class. To reduce the noises of the data, which was collected from the Internet, outlier images were detected and eliminated through a one-class SVM trained with deep convolutional features. We simultaneously trained a dish identifier, a cooking method recognizer, and a multi-label ingredient detector. They share a few low-level layers in the deep network architecture. The proposed framework shows higher accuracy than traditional method with handcrafted features, and the cooking method recognizer and ingredient detector can be applied to dishes which are not included in the training dataset to provide reference information for users.

关键词： multi-task learning convolutional neural network food recognition machine learning