检索结果-内蒙古大学图书馆

ieee/CVF conference on Computer Vision and Pattern Recognition (CVPR)

作者： Liu, Lei Chen, Yuze Yan, Junchi Zheng, Yinqiang Shanghai Jiao Tong Univ Dept CSE Shanghai Peoples R China Shanghai Jiao Tong Univ MoE Key Lab Artificial Intelligence Shanghai Peoples R China Univ Tokyo Tokyo Japan

ISBN: (数字)9781665469463

ISBN: (纸本)9781665469463

The industry practice for night video surveillance is to use auxiliary near-infrared (NIR) LEDs, usually centered at 850nm or 940nm, for scene illumination. NIR LEDs are used to save power consumption while hiding the surveillance coverage area from naked human eyes. The captured images are almost monochromatic, and visual color and texture tend to disappear, which hinders human and machine perception. A few existing studies have tried to convert such NIR images to RGB images through deep learning, which can not provide satisfying results, nor generalize well beyond the training dataset. In this paper, we aim to break the fundamental restrictions on reliable NIR-to-RGB (NIR2RGB) translation by examining the imaging mechanism of single-chip silicon-based RGB cameras under NIR illuminations, and propose to retrieve the optimal LED multiplexing via deep learning. Experimental results show that this translation task can be significantly improved by properly multiplexing NIR LEDs close to the visible spectral range than using 850nm and 940nm LEDs.

关键词： Multiplexing Industries Deep learning Training visualization Lighting Light emitting diodes

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Hierarchical Semantic Segmentation with Multiview Cosegmentation and Clustering Transformers

Unsupervised Hierarchical Semantic Segmentation with Multivi...

引用

ieee/CVF conference on Computer Vision and Pattern Recognition (CVPR)

作者： Ke, Tsung-Wei Hwang, Jyh-Jing Guo, Yunhui Wang, Xudong Yu, Stella X. Univ Calif Berkeley ICSI Berkeley CA 94704 USA

ISBN: (数字)9781665469463

ISBN: (纸本)9781665469463

Unsupervised semantic segmentation aims to discover groupings within and across images that capture object-and view-invariance of a category without external supervision. Grouping naturally has levels of granularity, creating ambiguity in unsupervised segmentation. Existing methods avoid this ambiguity and treat it as a factor outside modeling, whereas we embrace it and desire hierarchical grouping consistency for unsupervised segmentation. We approach unsupervised segmentation as a pixel-wise feature learning problem. Our idea is that a good representation shall reveal not just a particular level of grouping, but any level of grouping in a consistent and predictable manner. We enforce spatial consistency of grouping and bootstrap feature learning with co-segmentation among multiple views of the same image, and enforce semantic consistency across the grouping hierarchy with clustering transformers between coarse- and fine-grained features. We deliver the first data-driven unsupervised hierarchical semantic segmentation method called Hierarchical Segment Grouping (HSG). Capturing visual similarity and statistical co-occurrences, HSG also outperforms existing unsupervised segmentation methods by a large margin on five major object- and scene-centric benchmarks.

关键词： Representation learning Image segmentation visualization Computer vision Semantics Benchmark testing Transformers

来源：评论

学校读者我要写书评

暂无评论

Sound and Visual Representation Learning with Multiple Pretraining Tasks

Sound and Visual Representation Learning with Multiple Pretr...

引用

ieee/CVF conference on Computer Vision and Pattern Recognition (CVPR)

作者： Vasudevan, Arun Balajee Dai, Dengxin Van Gool, Luc Swiss Fed Inst Technol Zurich Switzerland MPI Informat Saarbrucken Germany Katholieke Univ Leuven Leuven Belgium

ISBN: (数字)9781665469463

ISBN: (纸本)9781665469463

Different self-supervised tasks (SSL) reveal different features from the data. The learned feature representations can exhibit different performance for each downstream task. In this light, this work aims to combine Multiple SSL tasks (Multi-SSL) that generalizes well for all downstream tasks. For this study, we investigate binaural sounds and image data. For binaural sounds, we propose three SSL tasks namely, spatial alignment, temporal synchronization of foreground objects and binaural sounds and temporal gap prediction. We investigate several approaches of Multi-SSL and give insights into the downstream task performance on video retrieval, spatial sound super resolution, and semantic prediction using OmniAudio dataset. Our experiments on binaural sound representations demonstrate that Multi-SSL via incremental learning (IL) of SSL tasks outperforms single SSL task models and fully supervised models in the downstream task performance. As a check of applicability on other modalities, we also formulate our Multi-SSL models for image representation learning and we use the recently proposed SSL tasks, MoCov2 and DenseCL. Here, Multi-SSL surpasses recent methods such as MoCov2, DenseCL and DetCo by 2.06%, 3.27% and 1.19% on VOC07 classification and +2.83, +1.56 and +1.61 AP on COCO detection.

关键词： Representation learning visualization Semantics Image representation Predictive models Market research Pattern recognition

来源：评论

学校读者我要写书评

暂无评论

LONGITUDINAL CHANGES IN RESTING STATE FMRI SPECTRA IN CHILDREN 44

LONGITUDINAL CHANGES IN RESTING STATE FMRI SPECTRA IN CHILDR...

引用

44th Annual International conference of the ieee-Engineering-in-Medicine-and-Biology-Society (EMBC)

作者： Agcaoglu, Oktay Wilson, Tony W. Wang, Yu-Ping Stephen, Julia M. Calhoun, Vince Emory Univ Georgia State Univ Triinst Ctr Translat Res Neuroimaging & Data Sci Georgia Inst Technol 55 Pk Pl NE18th Floor Atlanta GA 30303 USA Boys Town Natl Res Hosp Inst Human Neurosci 378 Bucher Boys Town NE 68010 USA Tulane Univ Dept Biomed Engn 6823 St Charles Ave New Orleans LA 70118 USA Tulane Univ Dept Global Biostat & Data Sci 1440 Canal St New Orleans LA 70112 USA Mind Res Network 1101 Yale Blvd NE Albuquerque NM 87106 USA

ISBN: (数字)9781728127828

ISBN: (纸本)9781728127828

Longitudinal studies can provide more precise measure of brain development, as they focus on within-subject variability, as opposed to cross-sectional studies. In this study, we track longitudinal changes in resting state fMRI data using spectrum of time-courses generated via group independent component analysis (gICA), in a multi time point dataset containing healthy children 8-18 years old, collected on both eyes open and eyes closed resting state conditions.

关键词： Pediatrics visualization Mental disorders Focusing Independent component analysis Biomarkers Functional magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

Research on Fire Escape System Based on Virtual Reality Technology 5

Research on Fire Escape System Based on Virtual Reality Tech...

引用

5th ieee International conference on Electronics Technology, ICET 2022

作者： Lei, Chuanjie Ke, Siyue Pei, Zhangfei Shao, Wenbo Liu, Shanshan Wang, Yan Wuhan Business University School of Information Engineering Wuhan China

ISBN: (纸本)9781665485081

Virtual reality technology mainly provides participants with a realistic simulation experience by using computer simulation systems to simulate the external environment. The fire escape system based on virtual reality technology has the characteristics of immersion, interactivity, conceivability, and 3D visualization, which is conducive to improving the fire safety awareness of the experiencer and enhancing the authenticity of fire drills. This paper first introduced the background of the fire escape system, analyzed the characteristics of the fire escape system technology, focusing on particle collision and sensory impact, and the application of relevant technologies to virtual reality technology applied to the anti-escape system, through scene restoration, scenario reproduction and other ways to be able to optimize the firefighting and rescue disposal tactical program, to achieve the interactive function with the virtual scene, to provide a scientific plan to deal with similar cases in the future. Finally, a comparison is made between the conventional fire escape training method and the method proposed in this paper. The method proposed in this paper is more favorable to enhance the authenticity and experience of the fire escape system for the experiencer. © 2022 ieee.

关键词： Virtual reality

来源：评论

学校读者我要写书评

暂无评论

How EvoStreets Are Observed in Three-Dimensional and Virtual Reality Environments

How EvoStreets Are Observed in Three-Dimensional and Virtual...

引用

Fachtagung des GI-Fachbereichs Softwaretechnik, Software Engineering 2022 - conference of the GI Software Engineering Section, Software Engineering 2022

作者： Koschke, Rainer Steinbeck, Marcel Universität Bremen AG Softwaretechnik Bibliothekstrasse 5 Bremen28359 Germany

ISBN: (纸本)9783885797142

We present our paper published in the proceedings of the 27th ieee International conference on Software Analysis, Evolution and Reengineering (SANER). In an experiment in which 34 participants had to analyze cloning in existing software systems using the EvoStreets visualization, we found indications that the movement patterns of the participants differ depending on whether they are in a 2.5D (pseudo-3D;monitor with keyboard and mouse) or virtual reality (VR) environment. In a follow-up study, which we present in this work, we analyzed the results of this experiment in more details, to examine whether not only movement is affected by these environments, but also the way how EvoStreets are observed. Beyond that, the paper proposes six visualization and user interaction concepts that are specific to the kind of environment. © 2022 Gesellschaft fur Informatik (GI). All rights reserved.

关键词： visualization

来源：评论

学校读者我要写书评

暂无评论

Auto QA : The Question Is Not Only What, but Also Where

Auto QA : The Question Is Not Only What, but Also Where

引用

22nd ieee/CVF Winter conference on Applications of Computer Vision (WACV)

作者： Kumar, Sumit Patro, Badri N. Namboodiri, Vinay P. IIT Kanpur Kanpur Uttar Pradesh India Univ Bath Bath Avon England Katholieke Univ Leuven Leuven Belgium

ISBN: (纸本)9781665458245

Visual Question Answering can be a functionally relevant task if purposed as such. In this paper, we aim to investigate and evaluate its efficacy in terms of localization-based question answering. We do this specifically in the context of autonomous driving where this functionality is important. To achieve our aim, we provide a new dataset, Auto-QA. Our new dataset is built over the Argoverse dataset and provides a truly multi-modal setting with seven views per frame and point-cloud LIDAR data being available for answering a localization-based question. We contribute localized attention adaptations of most popular VQA baselines and evaluate them on this task. We also provide joint point-cloud and image-based baselines that perform well on this task. An additional evaluation that we perform is to analyse whether the attention module is accurate or not for the image-based VQA baselines. To summarize, through this work we thoroughly analyze the localization abilities through visual question answering for autonomous driving and provide a new benchmark task for the same. Our best joint baseline model achieves a useful 74.8% accuracy on this task.

关键词： Location awareness visualization Computer vision Laser radar conferences Semantics Sensor systems

来源：评论

学校读者我要写书评

暂无评论

Object Localization Assistive System Based on CV and Vibrotactile Encoding 44

Object Localization Assistive System Based on CV and Vibrota...

引用

44th Annual International conference of the ieee-Engineering-in-Medicine-and-Biology-Society (EMBC)

作者： Wei, Zhikai Song, Aiguo Hu, Xuhui Southeast Univ Sch Instrument Sci & Engn State Key Lab Bioelect Nanjing 210096 Jiangsu Peoples R China Southeast Univ Sch Instrument Sci & Engn Jiangsu Key Lab Remote Measurement & Control Nanjing 210096 Jiangsu Peoples R China

ISBN: (数字)9781728127828

ISBN: (纸本)9781728127828

Intelligent assistive systems can navigate blind people, but most of them could only give non-intuitive cues or inefficient guidance. Based on computer vision and vibrotactile encoding, this paper presents an interactive system that provides blind people with intuitive spatial cognition. Different from the traditional auditory feedback strategy based on speech cues, this paper firstly introduces a vibration-encoded feedback method that leverages the haptic neural pathway and enables the users to interact with objects other than manipulating an assistance device. Based on this strategy, a wearable visual module based on an RGB-D camera is adopted for 3D spatial object localization, which contributes to accurate perception and quick object localization in the real environment. The experimental results on target blind individuals indicate that vibrotactile feedback reduces the task completion time by over 25% compared with the mainstream voice prompt feedback scheme. The proposed object localization system provides a more intuitive spatial navigation and comfortable wearability for blindness assistance.

关键词： Location awareness visualization Three-dimensional displays Navigation Interactive systems Blindness Neural pathways

来源：评论

学校读者我要写书评

暂无评论

Six methods for transforming layered hypergraphs to apply layered graph layout algorithms

引用

COMPUTER GRAPHICS FORUM 2022年第3期41卷 259-270页

作者： Di Bartolomeo, Sara Pister, Alexis Buono, Paolo Plaisant, Catherine Dunne, Cody Fekete, Jean-Daniel Univ Paris Saclay INRIA CNRS Orsay France Northeastern Univ Boston MA 02115 USA Univ Maryland College Pk MD 20742 USA Univ Bari Bari Italy

Hypergraphs are a generalization of graphs in which edges (hyperedges) can connect more than two vertices-as opposed to ordinary graphs where edges involve only two vertices. Hypergraphs are a fairly common data structure but there is little consensus on how to visualize them. To optimize a hypergraph drawing for readability, we need a layout algorithm. Common graph layout algorithms only consider ordinary graphs and do not take hyperedges into account. We focus on layered hypergraphs, a particular class of hypergraphs that, like layered graphs, assigns every vertex to a layer, and the vertices in a layer are drawn aligned on a linear axis with the axes arranged in parallel. In this paper, we propose a general method to apply layered graph layout algorithms to layered hypergraphs. We introduce six different transformations for layered hypergraphs. The choice of transformation affects the subsequent graph layout algorithm in terms of computational performance and readability of the results. Thus, we perform a comparative evaluation of these transformations in terms of number of crossings, edge length, and impact on performance. We also provide two case studies showing how our transformations can be applied to real-life use cases. A copy of this paper with all appendices and supplemental material is available at ***/grvwu.

关键词： CCS Concepts • Human-centered computing → Graph drawings visualization theory concepts and paradigms

来源：评论

学校读者我要写书评

暂无评论

Retrieval of small and occluded objects in road scenes based on improved YOLOv5 algorithm 8

Retrieval of small and occluded objects in road scenes based...

引用

8th International conference on Intelligent Computing and Signal Processing, ICSP 2023

作者： Xiong, Xuhui Ran, Dongni Zhang, Tao Hubei Normal University College of Computer and Information Engineering Hubei Huangshi China

ISBN: (纸本)9798350302455

The traffic volume in road traffic affects our travel and safety. Maintaining smooth traffic and real-time monitoring and unblocking of congested vehicles during peak hours is crucial for our daily lives. Most of the objects in existing traffic road scenes are occluded and mostly small objects. This paper proposes an improved YOLOv5 method for road scenes with certain occlusion and small objects and applies it to video object retrieval. Firstly, the SimAM attention mechanism is introduced in the YOLOv5 header to improve the detection accuracy of small objects without introducing additional parameters. In addition, using DIoU- NMS replaces NMS as a predictive frame filtering algorithm to improve the problem of detecting object loss in the case of multiple overlapping objects. Then, the improved YOLOv5 model is introduced with structures such as Ghost and RepVGG to achieve the effect of a lightweight model. Finally, this article proposes to combine the improved YOLOv5 model with CLIP for video object retrieval, expanding the alignment ability between text description and the video frame where the object is located to the region level. The experimental results on the VisDrone2019 dataset show that the mAP0.5 of the improved YOLOv5 is 37.3%, which is 4% higher than the original YOLOv5. The lightweight model reduces the number of parameters by 43.5% and FLOPs by 51.7% while maintaining almost the same accuracy. The visualization results show that the combination of YOLOv5 and CLIP binding energy better retrieves the object in the road video. © 2023 ieee.

关键词： Binding energy

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：