检索结果-内蒙古大学图书馆

29th British Machine Vision Conference, BMVC 2018

作者： Mahadevan, Sabarinath Voigtlaender, Paul Leibe, Bastian Computer Vision Group Visual Computing Institute RWTH Aachen University Germany

Deep learning requires large amounts of training data to be effective. For the task of object segmentation, manually labelling data is very expensive, and hence interactive methods are needed. Following recent approaches, we develop an interactive object segmentation system which uses user input in the form of clicks as the input to a convolutional network. While previous methods use heuristic click sampling strategies to emulate user clicks during training, we propose a new iterative training strategy. During training, we iteratively add clicks based on the errors of the currently predicted segmentation. We show that our iterative training strategy together with additional improvements to the network architecture results in improved results over the state-of-the-art. © 2018. The copyright of this document resides with its authors.

关键词： Network architecture

来源：评论

学校读者我要写书评

暂无评论

User Interface Design for Disabled People Under the Influence of Time, Efficiency and Costs

User Interface Design for Disabled People Under the Influenc...

引用

作者： Rezaei, Yashar Abbasalizadeh Heisenberg, Gernot Heiden, Wolfgang Institute of Visual Computing Bonn-Rhein-Sieg University of Applied Sciences Germany

Applications being designed for disabled people so far are showing three main issues: specific target user group, specialized user interface (UI) and interdependence problem. In addition, three essential criteria do also affect the application's usability, namely time, efficiency and costs. In order to overcome these problems, we propose a different perspective of User-Centered Design (UCD) by dividing and analyzing the UI architecture design process over three interdependent spaces: User, Need and application. Finally we provide the reader with an algorithmic guideline towards minimizing the interdependence issue between interaction modalities. © Springer International Publishing Switzerland 2014.

关键词： User centered design

来源：评论

学校读者我要写书评

暂无评论

Shot Boundary Detection for Automatic Video Analysis of Historical Films 2nd

Shot Boundary Detection for Automatic Video Analysis of Hist...

引用

2nd International Workshop on Recent Advances in Digital Security: Biometrics and Forensics, BioFor 2019, 1st International Workshop on Pattern Recognition for Cultural Heritage, PatReCH 2019, 1st International Workshop eHealth in the Big Data and Deep Learning Era, e-BADLE 2019, International Workshop on Deep Understanding Shopper Behaviors and Interactions in Intelligent Retail Environments, DEEPRETAIL 2019 and Industrial session held at the 20th International Conference on Image Analysis and Processing, ICIAP 2019

作者： Helm, Daniel Kampel, Martin Institute of Visual Computing and Human-Centered Technology TU Wien Vienna Austria

ISBN: (纸本)9783030307530

In automatic video content analysis and film preservation, Shot Boundary Detection (SBD) is a fundamental pre-processing step. While previous research focuses on detecting Abrupt Transitions (AT) as well as Gradual Transitions (GT) in different video genres such as sports movies or news clips only few studies investigate in the detection of shot transitions in historical footage. The main aim of this paper is to create an SBD mechanism inspired by state-of-the-art algorithms which is applied and evaluated on a self-generated historical dataset as well as a publicly available dataset called Clipshots. Therefore, a three-stage pipeline is introduced consisting of a Candidate Frame Range Selection based on the network DeepSBD, Extraction of Convolutional Neural Network (CNN) Features and Similarity Calculation. A combination of pre-trained backbone CNNs such as ResNet, VGG19 and SqueezeNet with different similarity metrics like Cosine Similarity and Euclidean Distance are used and evaluated. The outcome of this paper displays that the proposed algorithm reaches promising results on detecting ATs in historical videos without the need of complex optimization and re-training processes. Furthermore, it points out the main challenges concerning historical footage such as damaged film reels, scratches or splices. The results of this paper contribute a significant base for future research on automatic video analysis of historical videos. © 2019, Springer Nature Switzerland AG.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

OAFuser: Toward Omni-Aperture Fusion for Light Field Semantic Segmentation

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第12期5卷 6225-6239页

作者： Teng, Fei Zhang, Jiaming Peng, Kunyu Wang, Yaonan Stiefelhagen, Rainer Yang, Kailun Hunan University School of Robotics National Engineering Research Center of Robot Visual Perception and Control Technology Changsha410082 China Karlsruhe Institute of Technology Institute for Anthropomatics and Robotics Karlsruhe76131 Germany Institute for Visual Computing ETH Zurich Zurich8092 Switzerland

Light field cameras are capable of capturing intricate angular and spatial details. This allows for acquiring complex light patterns and details from multiple angles, significantly enhancing the precision of image semantic segmentation. However, two significant issues arise: 1) The extensive angular information of light field cameras contains a large amount of redundant data, which is overwhelming for the limited hardware resources of intelligent agents. 2) A relative displacement difference exists in the data collected by different microlenses. To address these issues, we propose an omni-aperture fusion model (OAFuser) that leverages dense context from the central view and extracts the angular information from subaperture images to generate semantically consistent results. To simultaneously streamline the redundant information from the light field cameras and avoid feature loss during network propagation, we present a simple yet very effective subaperture fusion module (SAFM). This module efficiently embeds subaperture images in angular features, allowing the network to process each subaperture image with a minimal computational demand of only (∼1\rm GFlops). Furthermore, to address the mismatched spatial information across viewpoints, we present a center angular rectification module (CARM) to realize feature resorting and prevent feature occlusion caused by misalignment. The proposed OAFuser achieves state-of-the-art performance on four UrbanLF datasets in terms of all evaluation metrics and sets a new record of 84.93% in mIoU on the UrbanLF-Real Extended dataset, with a gain of +3.69%. The source code for OAFuser is available at https://***/FeiBryantkit/OAFuser. © 2020 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Networked displays for VR applications: Display as a service 22

Networked displays for VR applications: Display as a service

引用

JVRC 2012: Joint Virtual Reality Conference of 22nd International Conference on Artificial Reality and Telexistence, ICAT 2012 and 18th Eurographics Symposium on Virtual Environments, EGVE 2012 and 9th EuroVR Conference, EuroVR 2012

作者： Löffler, Alexander Pica, Luciano Hoffmann, Hilko Slusallek, Philipp Intel Visual Computing Institute Saarbrücken Germany Saarbrücken Germany

ISBN: (纸本)9783905674408

Stereoscopic Liquid Crystal Displays (LCDs) in a tiled setup, so-called display walls, are rising as a replacement for the classic projection-based systems for Virtual Reality (VR) applications. They have numerous benefits over projectors, the only drawback being their maximum size, which is why VR applications usually resort to using tiled display walls. Problems of display walls are the obvious bezels between single displays making up the wall and, most importantly, the complicated pipeline to display synchronized content across all participating screens. This becomes especially crucial when we are dealing with active-stereo content, where precisely timed display of the left and right stereo channels across the entire display area is essential. Usually, these scenarios require a variety of expensive, specialized hardware, which makes it difficult for such wall setups to spread more widely. In this paper, we present our service-oriented architecture Display as a Service (DaaS), which uses a virtualization approach to shift the problem of pixel distribution from specialized hardware to a generic software. DaaS provides network-transparent virtual framebuffers (VFBs) for pixel-producing applications to write into and virtual displays (VDs), which potentially span multiple physical displays making up a display wall, to present generated pixels on. Our architecture assumes network-enabled displays with integrated processing capabilities, such that all communication for pixel transport and synchronization between VFBs and VDs can happen entirely over IP networking using standard video streaming and Internet protocols. We show the feasibility of our approach in a heterogeneous use case scenario, evaluate latency and synchronization accuracy, and give an outlook for more potential applications in the field of VR. © The Eurographics Association 2012.

关键词： Virtual reality

来源：评论

学校读者我要写书评

暂无评论

You get by with a little help:The effects of variable guidance degrees on performance and mental state

引用

visual Informatics 2019年第4期3卷 177-191页

作者： Davide Ceneda Theresia Gschwandtner Silvia Miksch TU Wien Faculty of InformaticsInstitute of Visual Computing&Human-Centered TechnologyFavoritenstrasse 9-11/193A-1040 ViennaAustria

Since it can be challenging for users to effectively utilize interactive visualizations,guidance is usually provided to assist users in solving *** is mentioned as an effective mean to overcome stall situations occurring during the ***,the effectiveness of a peculiar guidance solution usually varies for different analysis *** same guidance may have different effects on users with(1)different levels of *** choice of the appropriate(2)degree of guidance and the type of(3)task under consideration also affect the positive or negative outcome of providing *** these three factors,we conducted a user study to investigate the effectiveness of variable degrees of guidance with respect to the user’s previous knowledge in different analysis *** results shed light on the appropriateness of certain degrees of guidance in relation to different tasks,and the overall influence of guidance on the analysis outcome in terms of user’s mental state and analysis performance.

关键词： Guidance User study Knowledge Trust Mixed-initiative visual data analysis

来源：评论

学校读者我要写书评

暂无评论

Real-time Traffic Cone Detection for Autonomous Vehicle 34

Real-time Traffic Cone Detection for Autonomous Vehicle

引用

第三十四届中国控制会议

作者： HUANG Yong XUE Jianru Visual Cognitive Computing and Intelligent Vehicle Lab Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University

ISBN: (纸本)9781467374439

Traffic signs recognition is a basic task for autonomous vehicle. Among the numerous traffic signs, traffic cone is a very important mark used to guide cars where to go. A method based on vision and radar sensors information fusion is proposed to detect traffic cone in this paper. The algorithm mainly includes two parts: finding where the obstacle is in the image and recognizing whether it is a cone. We use homography to calibrate camera and radar, from which the radar data can be mapped on the image and a small corresponding image patch can be easily cutout. Then, a method based on contour feature called chamfer matching is used to determine whether the obstacle in the image patch is a cone. The approach has been tested on our autonomous vehicle, which shows it can guarantee both effectiveness and instantaneity.

关键词： traffic cone,autonomous vehicle,real time,calibration,chamfer matching

来源：评论

学校读者我要写书评

暂无评论

A Multi-Feature Fusion Based Traffic Light Recognition Algorithm for Intelligent Vehicles 33

A Multi-Feature Fusion Based Traffic Light Recognition Algor...

引用

第三十三届中国控制会议

作者： ZHANG Yue XUE Jianru ZHANG Geng ZHANG Yingwei ZHENG Nanning Visual Cognitive Computing and Intelligent Vehicle Lab Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University

ISBN: (纸本)9781479947249

Traffic light recognition is a key technology for intelligent vehicles and Advanced Driver Assistance Systems(ADAS).This paper proposes a multi-feature fusion based real-time traffic light recognition algorithm for intelligent *** the region of interest determined by the vanishing line,technologies including color segmentation,blob detection,and structural feature extraction are employed individually to obtain a set of candidate locations.A fusion algorithm is developed to integrate these results and compute a score for all these possible locations of traffic *** score of each candidate denotes its probability of being a traffic *** final detection is achieved by fusing its score with temporal and geographic *** experimental results on a real intelligent vehicle show that the proposed algorithm is effective and efficient.

关键词： Traffic Light,intelligent vehicle,multiple features,geographic information

来源：评论

学校读者我要写书评

暂无评论

Inter-hospital transferability of AI: A case study on phase recognition in cholecystectomy

引用

Computers in Biology and Medicine 2025年第PT B期192卷 110235-110235页

作者： Renz-Kiefel, Lasse Lünse, Sebastian Mantke, Rene Eisert, Peter Hilsmann, Anna Wisotzky, Eric L. Fraunhofer Heinrich-Hertz-Institute Vision & Imaging Technologies Berlin Germany Brandenburg Medical School Department of Surgery University Hospital Brandenburg Germany Faculty of Health Sciences Joint Faculty of the Brandenburg University of Technology Cottbus - Senftenberg the Brandenburg Medical School Theodor Fontane and the University of Potsdam & Department of Surgery University Hospital Brandenburg Germany Humboldt-University Berlin Visual Computing Berlin Germany Rostock University Medical Center Klinik und Poliklinik für Hals-Nasen-Ohrenheilkunde Kopf- und Halschirurgie "Otto Körner" Rostock Germany

Background: Identifying surgical phases is a crucial component of surgical workflow analysis, facilitating the automated evaluation of surgical procedures’ performance and efficiency. A significant challenge in developing neural networks for surgical phase recognition lies in the scarcity of training data and the large variation in surgical techniques among surgeons. Consequently, it is imperative for these networks to possess generalization capabilities across diverse datasets. In this paper, we analyze the transferability of trained phase recognition models, using cholecystectomy as a case study. Methods: We employed datasets comprising 104 publicly available surgeries from three different centers for training and conducted multiple experiments using 21 videos of surgeries we recorded ourselves for evaluation. A two-stage deep learning architecture was employed, using a ResNet50 backbone followed by a multi-stage Temporal Convolutional Network (MS-TCN). Several experiments were conducted, including training solely on MHB data, training exclusively on public data, and training on a combination of both with an additional fine-tuning approach. Results: Models trained solely on MHB data achieved an accuracy of approximately 79.7%, while those trained on public data alone performed significantly worse when applied to MHB data. The best performance was obtained by retraining on a combined dataset. The results indicate that it is possible to transfer models to new environments (operating rooms or clinics) and surgeons by using public data, and incorporating site-specific data improves model transferability. Conclusion: The results demonstrate that leveraging diverse training data, including institution-specific videos, is crucial to develop robust and transferable AI models for surgical phase recognition, thereby enhancing the potential of automated decision-support systems across different clinical environments. © 2025 The Authors

关键词： Laparoscopy

来源：评论

学校读者我要写书评

暂无评论

Saccade landing position prediction for gaze-contingent rendering

Saccade landing position prediction for gaze-contingent rend...

引用

ACM SIGGRAPH 2017

作者： Arabadzhiyska, Elena Tursun, Okan Tarhan Myszkowski, Karol Seidel, Hans-Peter Didyk, Piotr Saarland University MMCI Intel Visual Computing Institute Germany MPI Informatik Germany

Gaze-contingent rendering shows promise in improving perceived quality by providing a better match between image quality and the human visual system requirements. For example, information about fixation allows rendering quality to be reduced in peripheral vision, and the additional resources can be used to improve the quality in the foveal region. Gaze-contingent rendering can also be used to compensate for certain limitations of display devices, such as reduced dynamic range or lack of accommodation cues. Despite this potential and the recent drop in the prices of eye trackers, the adoption of such solutions is hampered by system latency which leads to a mismatch between image quality and the actual gaze location. This is especially apparent during fast saccadic movements when the information about gaze location is significantly delayed, and the quality mismatch can be noticed. To address this problem, we suggest a new way of updating images in gaze-contingent rendering during saccades. Instead of rendering according to the current gaze position, our technique predicts where the saccade is likely to end and provides an image for the new fixation location as soon as the prediction is available. While the quality mismatch during the saccade remains unnoticed due to saccadic suppression, a correct image for the new fixation is provided before the fixation is established. This paper describes the derivation of a model for predicting saccade landing positions and demonstrates how it can be used in the context of gaze-contingent rendering to reduce the influence of system latency on the perceived quality. The technique is validated in a series of experiments for various combinations of display frame rate and eye-tracker sampling rate. © 2017 Copyright held by the owner/author(s).

关键词： Forecasting

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：