检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Yun, Stone Wong, Alexander Vision and Image Processing Group University of Waterloo Waterloo Artificial Intelligence Institute Canada

Deep convolutional neural network (CNN) training via iterative optimization has had incredible success in finding optimal parameters. However, modern CNN architectures often contain millions of parameters. Thus, any given model for a single architecture resides in a massive parameter space. Models with similar loss could have drastically different characteristics such as adversarial robustness, generalizability, and quantization robustness. For deep learning on the edge, quantization robustness is often crucial. Finding a model that is quantization-robust can sometimes require significant efforts. Recent works using Graph Hypernetworks (GHN) have shown remarkable performance predicting high-performant parameters of varying CNN architectures. Inspired by these successes, we wonder if the graph representations of GHN-2 can be leveraged to predict quantization-robust parameters as well, which we call GHN-Q. We conduct the first-ever study exploring the use of graph hypernetworks for predicting parameters of unseen quantized CNN architectures. We focus on a reduced CNN search space and find that GHN-Q can in fact predict quantization-robust parameters for various 8-bit quantized CNNs. Decent quantized accuracies are observed even with 4-bit quantization despite GHN-Q not being trained on it. Quantized finetuning of GHN-Q at lower bitwidths may bring further improvements and is currently being explored. © 2022, CC BY-SA.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

CHASE: Robust Visual Tracking via Cell-Level Differentiable Neural Architecture Search 32

CHASE: Robust Visual Tracking via Cell-Level Differentiable ...

引用

32nd British Machine vision Conference, BMVC 2021

作者： Marvasti-Zadeh, Seyed Mojtaba Khaghani, Javad Cheng, Li Ghanei-Yakhdan, Hossein Kasaei, Shohreh Vision and Learning Lab University of Alberta Edmonton Canada Digital Image & Video Processing Lab Yazd University Yazd Iran Image Processing Lab Sharif University of Technology Tehran Iran

A strong visual object tracker nowadays relies on its well-crafted modules, which typically consist of manually-designed network architectures to deliver high-quality tracking results. Not surprisingly, the manual design process becomes a particularly challenging barrier, as it demands sufficient prior experience, enormous effort, intuition, and perhaps some good luck. Meanwhile, neural architecture search has gaining grounds in practical applications as a promising method in tackling the issue of automated search of feasible network structures. In this work, we propose a novel cell-level differentiable architecture search mechanism with early stopping to automate the network design of the tracking module, aiming to adapt backbone features to the objective of Siamese tracking networks during offline training. Besides, the proposed early stopping strategy avoids over-fitting and performance collapse problems leading to generalization improvement. The proposed approach is simple, efficient, and with no need to stack a series of modules to construct a network. Our approach is easy to be incorporated into existing trackers, which is empirically validated using different differentiable architecture search-based methods and tracking objectives. Extensive experimental evaluations demonstrate the superior performance of our approach over five commonly-used benchmarks. © 2021. 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

Hierarchical Attention Approach in Multimodal Emotion Recognition for Human Robot Interaction 36

Hierarchical Attention Approach in Multimodal Emotion Recogn...

引用

36th International Technical Conference on Circuits/Systems, Computers and Communications, ITC-CSCC 2021

作者： Abdullah, Muhammad Ahmad, Mobeen Han, Dongil Vision and Image Processing Lab Sejong University Department of Computer Engineering Seoul Korea Republic of

ISBN: (纸本)9781665435536

The ability to perceive human emotions is one of the key elements that may promise a natural, genuine and more reliable human robot interaction. Though emotional perception in human robot interaction has been challenged by many difficulties, the lack of contextual understanding is can be attributed as the biggest hurdle in this regard. Most of the literature refers to the datasets developed in controlled environment to validate the performance of their systems which happens to be really good. Still those systems are very far from achieving that kind of performance in real-life scenarios. In this paper a multimodal emotion recognition strategy is presented, that uses voice features in addition to the facial expressions to determine the emotional state of the user. A hierarchical attention layer is used to for feature fusion purpose. The final system is end-to-end trainable, multimodal approach makes it more resilient to the changes in environment. The achieved 76.3% accuracy for eNterface05 video dataset, which is higher than the any single modality approach in the comparison. © 2021 IEEE.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Is there progress in activity progress prediction?

Is there progress in activity progress prediction?

引用

International Conference on Computer vision Workshops (ICCV Workshops)

作者： Frans de Boer Jan C. van Gemert Jouke Dijkstra Silvia L. Pintea Computer Vision Lab Delft University of Technology Division of Image Processing (LKEB) Leiden University Medical Center

Activity progress prediction aims to estimate what percentage of an activity has been completed. Currently this is done with machine learning approaches, trained and evaluated on complicated and realistic video datasets. The videos in these datasets vary drastically in length and appearance. And some of the activities have unanticipated developments, making activity progression difficult to estimate. In this work, we examine the results obtained by existing progress prediction methods on these datasets. We find that current progress prediction methods seem not to extract useful visual information for the progress prediction task. Therefore, these methods fail to exceed simple frame-counting baselines. We design a precisely controlled dataset for activity progress prediction and on this synthetic dataset we show that the considered methods can make use of the visual information, when this directly relates to the progress prediction. We conclude that the progress prediction task is ill-posed on the currently used real-world datasets. Moreover, to fairly measure activity progression we advise to consider a, simple but effective, frame-counting baseline.

关键词：

来源：评论

学校读者我要写书评

暂无评论

COVIDx CT-3: A Large-scale, Multinational, Open-Source Benchmark Dataset for Computer-aided COVID-19 Screening from Chest CT images

arXiv

引用

arXiv 2022年

作者： Gunraj, Hayden Tuinstra, Tia Wong, Alexander Vision and Image Processing Lab University of Waterloo Canada Department of Systems Design Engineering University of Waterloo Canada Waterloo AI Institute University of Waterloo Canada DarwinAI Corp Canada

Computed tomography (CT) has been widely explored as a COVID-19 screening and assessment tool to complement RT-PCR testing. To assist radiologists with CT-based COVID-19 screening, a number of computer-aided systems have been proposed. However, many proposed systems are built using CT data which is limited in both quantity and diversity. Motivated to support efforts in the development of machine learning-driven screening systems, we introduce COVIDx CT-3, a large-scale multinational benchmark dataset for detection of COVID-19 cases from chest CT images. COVIDx CT-3 includes 431,205 CT slices from 6,068 patients across at least 17 countries, which to the best of our knowledge represents the largest, most diverse dataset of COVID-19 CT images in open-access form. Additionally, we examine the data diversity and potential biases of the COVIDx CT-3 dataset, finding that significant geographic and class imbalances remain despite efforts to curate data from a wide variety of sources. Copyright © 2022, The Authors. All rights reserved.

关键词： Large dataset

来源：评论

学校读者我要写书评

暂无评论

Tissue-Specific labelling using Temporal Feature Extraction in Photoacoustic Remote Sensing Microscopy

Tissue-Specific Labelling using Temporal Feature Extraction ...

引用

Clinical and Translational Biophotonics, Translational 2022

作者： Tweel, James Pellegrino, Nicholas Ecclestone, Benjamin R. Fieguth, Paul Reza, Parsin Haji PhotoMedicine Labs Systems Design Engineering University of Waterloo WaterlooONN2L 3G1 Canada Vision and Image Processing Lab Systems Design Engineering University of Waterloo WaterlooONN2L 3G1 Canada Illumisonics Inc. 22 King Street South WaterlooONN2J 1N8 Canada

Characteristic time-domain (TD) features are extracted from Photoacoustic Remote Sensing TD modulation using a novel temporal clustering algorithm. Features correspond to tissue-specific structures in unprocessed rese... 详细信息

ISBN: (纸本)9781557528209

关键词： Tissue

来源：评论

学校读者我要写书评

暂无评论

The Influence of Input Variable Selection on Deep Learning-Based Sea Ice Parameter Inversion from Multi-Sensor Satellite Data

The Influence of Input Variable Selection on Deep Learning-B...

引用

OCEANS

作者： Xinwei Chen Muhammed Patel Fernando J. Pena Cantu Jinman Park Javier Noa Turnes Mingzhe Jiang Linlin Xu K. Andrea Scott David A. Clausi Weimin Huang Department of System Design Engineering Vision and Image Processing Lab University of Waterloo Faculty of Engineering and Applied Science Memorial University of Newfoundland St. John's NL Canada

The automated mapping of Arctic sea ice using satellite data from various sources plays a critical role in maritime shipping and climate science research. Deep learning models, particularly convolutional neural networks (CNNs), have shown significant potential in generating more accurate and reliable automated sea ice maps from synthetic aperture radar (SAR) images. This study focuses on investigating the impact of integrating multi-source data on enhancing the performance of CNN-based sea ice mapping models. By employing a multitask U-Net architecture, the paper aims to estimate three essential sea ice parameters, namely sea ice concentration (SIC), stage of development (SOD), and floe size (FLOE), through different combinations of input data. The AI4Arctic Sea Ice Challenge dataset with multi-sensor satellite data is utilized for model training and evaluation. The experimental findings confirm the effectiveness of incorporating passive microwave data and spatial-temporal information in conjunction with SAR imagery, as well as feature redundancy removal, resulting in improved accuracy for sea ice parameter estimation.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Pixel-Level Sea Ice Concentration Retrieval from Sentinel-1 SAR imagery and Ancillary Data via Regional Loss Representations

Pixel-Level Sea Ice Concentration Retrieval from Sentinel-1 ...

引用

OCEANS

作者： Xinwei Chen Muhammed Patel Linlin Xu Yuhao Chen David A. Clausi K. Andrea Scott Weimin Huang School of Marine Science and Engineering South China University of Technology Department of System Design Engineering Vision and Image Processing Lab University of Waterloo Department of Mechanical & Mechatronics Engineering University of Waterloo Faculty of Engineering and Applied Science Memorial University of Newfoundland St. John's NL Canada

ISBN: (数字)9798350362077

ISBN: (纸本)9798350362084

The automated monitoring and high resolution mapping of Arctic sea ice concentration (SIC) from synthetic aperture radar (SAR) images are crucial for various purposes, such as tactical navigation. However, the available labels for training deep learning-based SIC estimation models are mostly in coarse spatial resolution. This study tackles this key issue by proposing a novel loss calculation algorithm during model training. Consequently, a U-Net-based model can produce fine-grained SIC estimates via direct learning from ice charts in low resolution. The proposed method is evaluated with the recently release AI4Arctic Sea Ice Challenge Dataset with more than 500 Sentinel-1 SAR imager and associated ice charts. Visual interpretation and numerical results demonstrate the outstanding performance of the proposed method in terms of SIC mapping resolution. The effectiveness of incorporating multi-source ancillary data in performance improvement is also validated.

关键词： Training Visualization Sentinel-1 Estimation Radar polarimetry Numerical models Spatial resolution

来源：评论

学校读者我要写书评

暂无评论

Quantization in Relative Gradient Angle Domain For Building Polygon Estimation

Quantization in Relative Gradient Angle Domain For Building ...

引用

International Conference on Pattern Recognition

作者： Yuhao Chen Yifan Wu Linlin Xu Alexander Wong Vision and Image Processing Lab (VIP) University of Waterloo Waterloo Ontario Canada

Building footprint extraction in remote sensing data benefits many important applications, such as urban planning and population estimation. Recently, rapid development of convolutional neural networks (CNNs) and open-sourced high resolution satellite building image datasets have pushed the performance boundary further for automated building extractions. However, CNN approaches often generate imprecise building morphologies including noisy edges and round corners. In this paper, we leverage the performance of CNNs, and propose a module that uses prior knowledge of building corners to create angular and concise building polygons from CNN segmentation outputs. We describe a new transform, Relative Gradient Angle Transform (RGA Transform) that converts object contours from time vs. space to time vs. angle. We propose a new shape descriptor, Boundary Orientation Relation Set (BORS), to describe angle relationship between edges in RGA domain, such as orthogonality and parallelism. Finally, we develop an energy minimization framework that makes use of the angle relationship in BORS to straighten edges and reconstruct sharp corners, and the resulting corners create a polygon. Experimental results demonstrate that our method refines CNN output from a rounded approximation to a more clear-cut angular shape of the building footprint.

关键词： Shape Buildings Urban planning Sociology Estimation Transforms Minimization

来源：评论

学校读者我要写书评

暂无评论

Text Detection & Recognition in the Wild for Robot Localization

arXiv

引用

arXiv 2022年

作者： Raisi, Zobeir Zelek, John Vision and Image Processing Lab Systems Design Engineering Department University of Waterloo WaterlooONN2L 3G1 Canada

Signage is everywhere and a robot should be able to take advantage of signs to help it localize (including Visual Place Recognition (VPR)) and map. Robust text detection & recognition in the wild is challenging due such factors as pose, irregular text, illumination and occlusion. We propose an end-to-end scene text spotting model that simultaneously outputs the text string and bounding boxes. This model is more suitable for VPR. Our central contribution is introducing utilizing an end-to-end scene text spotting framework to adequately capture the irregular and occluded text regions in different challenging places. To evaluate our proposed architecture’s performance for VPR, we conducted several experiments on the challenging Self-Collected Text Place (SCTP) benchmark dataset. The initial experimental results show that the proposed method outperforms the SOTA methods in terms of precision and recall when tested on this benchmark. © 2022, CC BY-NC-SA.

关键词： Robot applications

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：