检索结果-内蒙古大学图书馆

Dual-Level Masked Semantic Inference for Semi-Supervised Semantic Segmentation

IEEE Transactions on Multimedia 2025年

作者： Ma, Qiankun Zhang, Ziyao Qiao, Pengchong Wang, Yu Ji, Rongrong Liu, Chang Chen, Jie Peking University School of Electronic and Computer Engineering Shenzhen518055 China Chinese Academy of Sciences Paul C. Lauterbur Research Center for Biomedical Imaging Shenzhen Institutes of Advanced Technology Shenzhen518055 China University of the Chinese Academy of Sciences Beijing100049 China Peng Cheng Laboratory Shenzhen518055 China Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence School of Informatics Xiamen361005 China Tsinghua University Department of Automation Beijing100084 China

Semi-supervised semantic segmentation pursues a holistic pixel-wise understanding of unseen images with limited annotation. To this end, existing methods focus on regularizing per-pixel prediction consistency within unlabeled data, while rarely modeling contextual relationships. But in fact, contextual semantics can provide valuable clues for scene understanding like inner-object continuity and spatial relationships' causality. Thus, in this paper, we propose a Dual-level Masked Semantics Inference (DMSI) that takes the initiative to explicitly learn contextual relationships via enforcing our model to infer the semantics of a pixel according to its surrounding contexts. This allows our model to exhaust accurate semantics by incorporating inter-pixel context clues, further leading to comprehensive segmentation. Specifically, DMSI comprises two main components: 1) Dual-level mask consistency regularization (DMCR) that learns the ability of semantics inference by aligning the predictions of masked views with the prediction of the complete view. The masked views here come from both the image level and feature level, where our model captures low-level attributes and high-level representations respectively. 2) AdaMask that provides a proper mask position and ratio for each image, guiding our model to focus on semantic-rich regions while providing balanced training between hard and easy samples. Through learning the ability of semantic inferring, DMSI remarkably enhances the interaction between pixels, further progressively intensifying the understanding of semantics. Extensive experiments under various settings on Cityscapes and Pascal VOC 2012 show that DMSI achieves new state-of-the-art performances. Furthermore, analysis indicates that our method has superiority in mining inter-pixel semantic relationships and improving robustness facing noise corruption. © 1999-2012 IEEE.

关键词： Latent semantic analysis

来源：评论

学校读者我要写书评

暂无评论

Toward Efficient CNN Compression via Coordinate Descent Structure Search

Toward Efficient CNN Compression via Coordinate Descent Stru...

引用

2021 IEEE Conference on Telecommunications, Optics and Computer Science, TOCS 2021

作者： Zhang, Yuxin Ji, Rongrong Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen China

ISBN: (纸本)9781665424981

Despite the impressive performance, convolution neural networks (CNNs) are often pilloried for challenging deployment on edge devices due to huge computation burden and power consumption. As an effective tool to compress CNNs, filter pruning has attracted ever-increasing attention in recent years. However, most methods suffer from complex manual design, huge overhead of structure search and unstable compression rate. To address these problems, this paper proposes a novel end-To-end pruning method based on coordinate descent, termed CDPruner. Specifically, the proposed method models filter pruning as an independent optimization for the pruning rate of each layer. By assigning each layer a pruning sensitivity score, CDPruner selects the least sensitive layer to prune in an iterative manner and dynamically updates the sensitivity queue until the FLOPs reduction reaches the pre-given budget. Extensive experiments on CIFAR-10 and ImageNet-IK dataset demonstrate the superiority of CDPruner over many state-of-The-Arts. © 2021 IEEE.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Improving Occluded Human Pose Estimation Via Linked Joints 48

Improving Occluded Human Pose Estimation Via Linked Joints

引用

48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

作者： Ye, Suhang Hong, Zebo Zheng, Jiawen Zhang, Shengchuan Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence School of Informatics China

ISBN: (纸本)9781728163277

Keypoint heatmaps, which produce peak values of Gaussian distributions for individual human joints, are crucial components in 2D human pose estimation. However, existing merits using keypoint heatmaps are usually defeated in body occlusion, resulting in inaccurate joint predictions. We consider that the failure is mainly due to keypoint heatmaps' insufficiency for distinguishing the joints from two occluded bodies. Therefore, in this paper, we propose a method termed SkeletonMap (SMap), which introduces the prior knowledge of body structure to constrain relative connection of joints. As an extension of keypoint heatmaps, SMap can be efficiently plugged into existing 2D human pose estimation models with negligible increase in computational cost. Extensive experiments are conducted to show the effectiveness and generalization of SMap. Without bells and whistles, SMap brings a significant performance boost to the existing heatmap-based 2D human pose estimation models. On the MPII dataset, SMap improves SimpleBaseline (ResNet-152) from89.7 PCKh@0.5 to 90.4, and HRNet (W32) from 90.5 to 90.9. On the COCO dataset, SMap improves SimpleBaseline (ResNet-152) from 72.4 AP to 73.9. On the more challenging OCHuman dataset, SMap improves HRNet (W32) from 61.9 AP to 64.5, achieving 2.6 AP gains. We hope our simple and efficient approach will serve as a solid component for future research in 2D human pose estimation. © 2023 IEEE.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

Enhance Model Stealing Attack via Label Refining 7

Enhance Model Stealing Attack via Label Refining

引用

7th International Conference on Intelligent computing and Signal Processing, ICSP 2022

作者： Wang, Yixu Lin, Xianming School of Informatics Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen China

ISBN: (纸本)9781665478571

With machine learning models being increasingly deployed, model stealing attacks have raised an increasing interest. Extracting decision-based models is a more challenging task with the information of class similarity missing. In this paper, we propose a novel and effective model stealing method as Label Refining via Feature Distance (LRFD), to re-dig the class similarity. Specifically, since the information of class similarity can be represented by the distance between samples from different classes in the feature space, we design a soft label construction module inspired by the prototype learning, and transfer the knowledge in the soft label to the substitution model. Extensive experiments conducted on four widely-used datasets consistently demonstrate that our method yields a model with significantly greater functional similarity to the victim model. © 2022 IEEE.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

Carrying Out CNN Channel Pruning in a White Box

引用

IEEE Transactions on Neural Networks and Learning Systems 2023年第10期34卷 7946-7955页

作者： Zhang, Yuxin Lin, Mingbao Lin, Chia-Wen Chen, Jie Wu, Yongjian Tian, Yonghong Ji, Rongrong Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence School of Informatics Xiamen361005 China Youtu Laboratory Tencent Shanghai200233 China National Tsing Hua University Department of Electrical Engineering The Institute of Communications Engineering Hsinchu30013 Taiwan Peng Cheng Laboratory Shenzhen518000 China Peking University Department of Computer Science and Technology Beijing100871 China School of Informatics and the Institute of Artificial Intelligence Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen361005 China

Channel pruning has been long studied to compress convolutional neural networks (CNNs), which significantly reduces the overall computation. Prior works implement channel pruning in an unexplainable manner, which tends to reduce the final classification errors while failing to consider the internal influence of each channel. In this article, we conduct channel pruning in a white box. Through deep visualization of feature maps activated by different channels, we observe that different channels have a varying contribution to different categories in image classification. Inspired by this, we choose to preserve channels contributing to most categories. Specifically, to model the contribution of each channel to differentiating categories, we develop a class-wise mask for each channel, implemented in a dynamic training manner with respect to the input image's category. On the basis of the learned class-wise mask, we perform a global voting mechanism to remove channels with less category discrimination. Lastly, a fine-tuning process is conducted to recover the performance of the pruned model. To our best knowledge, it is the first time that CNN interpretability theory is considered to guide channel pruning. Extensive experiments on representative image classification tasks demonstrate the superiority of our White-Box over many state-of-the-arts (SOTAs). For instance, on CIFAR-10, it reduces 65.23% floating point operations per seconds (FLOPs) with even 0.62% accuracy improvement for ResNet-110. On ILSVRC-2012, White-Box achieves a 45.6% FLOP reduction with only a small loss of 0.83% in the top-1 accuracy for ResNet-50. Code is available at https://***/zyxxμWhite-Box. © 2012 IEEE.

关键词： Visualization

来源：评论

学校读者我要写书评

暂无评论

A Real-Time Global Inference Network for One-Stage Referring Expression Comprehension

引用

IEEE Transactions on Neural Networks and Learning Systems 2023年第1期34卷 134-143页

作者： Zhou, Yiyi Ji, Rongrong Luo, Gen Sun, Xiaoshuai Su, Jinsong Ding, Xinghao Lin, Chia-Wen Tian, Qi Xiamen University Media Analytics and Computing Laboratory Xiamen361005 China Xiamen University School of Informatics Informatics and Communication Engineering Department Xiamen361005 China School of Informatics Xiamen University Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen361005 China Xiamen University School of Informatics Xiamen361005 China National Tsing Hua University Department of Electrical Engineering Hsinchu30013 Taiwan Institute of Communications Engineering National Tsing Hua University Hsinchu30013 Taiwan Huawei Technologies Company Ltd. Cloud BU Shenzhen518129 China

Referring expression comprehension (REC) is an emerging research topic in computer vision, which refers to the detection of a target region in an image given a test description. Most existing REC methods follow a multistage pipeline, which is computationally expensive and greatly limits the applications of REC. In this article, we propose a one-stage model toward real-time REC, termed real-time global inference network (RealGIN). RealGIN addresses the issues of expression diversity and complexity of REC with two innovative designs: adaptive feature selection (AFS) and Global Attentive ReAsoNing (GARAN). Expression diversity concerns varying expression content, which includes information such as colors, attributes, locations, and fine-grained categories. To address this issue, AFS adaptively fuses features of different semantic levels to tackle the changes in expression content. In contrast, expression complexity concerns the complex relational conditions in expressions that are used to identify the referent. To this end, GARAN uses the textual feature as a pivot to collect expression-aware visual information from all regions and then diffuses this information back to each region, which provides sufficient context for modeling the relational conditions in expressions. On five benchmark datasets, i.e., RefCOCO, RefCOCO+, RefCOCOg, ReferIT, and Flickr30k, the proposed RealGIN outperforms most existing methods and achieves very competitive performances against the most advanced one, i.e., MAttNet. More importantly, under the same hardware, RealGIN can boost the processing speed by 10-20 times over the existing methods. © 2012 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Toward Efficient CNN Compression via Coordinate Descent Structure Search

Toward Efficient CNN Compression via Coordinate Descent Stru...

引用

Telecommunications, Optics and Computer Science (TOCS), IEEE Conference on

作者： Yuxin Zhang Rongrong Ji Media Analytics and Computing Laboratory School of Informatics Xiamen University Xiamen China

ISBN: (纸本)9781665424998

Despite the impressive performance, convolution neural networks (CNNs) are often pilloried for challenging deployment on edge devices due to huge computation burden and power consumption. As an effective tool to compress CNNs, filter pruning has attracted ever-increasing attention in recent years. However, most methods suffer from complex manual design, huge overhead of structure search and unstable compression rate. To address these problems, this paper proposes a novel end-to-end pruning method based on coordinate descent, termed CDPruner. Specifically, the proposed method models filter pruning as an independent optimization for the pruning rate of each layer. By assigning each layer a pruning sensitivity score, CDPruner selects the least sensitive layer to prune in an iterative manner and dynamically updates the sensitivity queue until the FLOPs reduction reaches the pre-given budget. Extensive experiments on CIFAR-10 and ImageNet-IK dataset demonstrate the superiority of CDPruner over many state-of-the-arts.

关键词： Optical filters Performance evaluation Sensitivity Power demand Neural networks Search problems Optics

来源：评论

学校读者我要写书评

暂无评论

Improving Occluded Human Pose Estimation Via Linked Joints

Improving Occluded Human Pose Estimation Via Linked Joints

引用

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

作者： Suhang Ye Zebo Hong Jiawen Zheng Shengchuan Zhang Department of Artificial Intelligence School of Informatics Media Analytics and Computing Laboratory Xiamen University China

Keypoint heatmaps, which produce peak values of Gaussian distributions for individual human joints, are crucial components in 2D human pose estimation. However, existing merits using keypoint heatmaps are usually defeated in body occlusion, resulting in inaccurate joint predictions. We consider that the failure is mainly due to keypoint heatmaps’ insufficiency for distinguishing the joints from two occluded bodies. Therefore, in this paper, we propose a method termed SkeletonMap (SMap), which introduces the prior knowledge of body structure to constrain relative connection of joints. As an extension of keypoint heatmaps, SMap can be efficiently plugged into existing 2D human pose estimation models with negligible increase in computational cost. Extensive experiments are conducted to show the effectiveness and generalization of SMap. Without bells and whistles, SMap brings a significant performance boost to the existing heatmap-based 2D human pose estimation models. On the MPII dataset, SMap improves SimpleBaseline (ResNet-152) from89.7 PCKh@0.5 to 90.4, and HRNet (W32) from 90.5 to 90.9. On the COCO dataset, SMap improves SimpleBaseline (ResNet-152) from 72.4 AP to 73.9. On the more challenging OCHuman dataset, SMap improves HRNet (W32) from 61.9 AP to 64.5, achieving 2.6 AP gains. We hope our simple and efficient approach will serve as a solid component for future research in 2D human pose estimation.

关键词： Heating systems Computational modeling Pose estimation Signal processing Gaussian distribution Benchmark testing Solids

来源：评论

学校读者我要写书评

暂无评论

SiMaN: Sign-to-Magnitude Network Binarization

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2023年第5期45卷 6277-6288页

作者： Lin, Mingbao Ji, Rongrong Xu, Zihan Zhang, Baochang Chao, Fei Lin, Chia-Wen Shao, Ling Xiamen University School of Informatics Media Analytics and Computing Laboratory Department of Artificial Intelligence Xiamen361005 China Tencent Youtu Lab Shanghai200233 China Zhongguancun Lab Beijing100190 China National Tsing Hua University Department of Electrical Engineering The Institute of Communications Engineering Hsinchu30013 Taiwan Terminus Group Beijing100026 China

Binary neural networks (BNNs) have attracted broad research interest due to their efficient storage and computational ability. Nevertheless, a significant challenge of BNNs lies in handling discrete constraints while ensuring bit entropy maximization, which typically makes their weight optimization very difficult. Existing methods relax the learning using the sign function, which simply encodes positive weights into +1s, and -1s otherwise. Alternatively, we formulate an angle alignment objective to constrain the weight binarization to {0,+1} to solve the challenge. In this article, we show that our weight binarization provides an analytical solution by encoding high-magnitude weights into +1s, and 0s otherwise. Therefore, a high-quality discrete solution is established in a computationally efficient manner without the sign function. We prove that the learned weights of binarized networks roughly follow a Laplacian distribution that does not allow entropy maximization, and further demonstrate that it can be effectively solved by simply removing the 2 regularization during network training. Our method, dubbed sign-to-magnitude network binarization (SiMaN), is evaluated on CIFAR-10 and ImageNet, demonstrating its superiority over the sign-based state-of-the-arts. Our source code, experimental settings, training logs and binary models are available at https://***/lmbxμSiMaN. © 1979-2012 IEEE.

关键词： Quantization (signal)

来源：评论

学校读者我要写书评

暂无评论

An Efficient Blur Kernel Estimation Method for Blind Image Super-Resolution

SSRN

引用

SSRN 2024年

作者： Xu, Yimin Gao, Nanxi Chao, Fei Ji, Rongrong Media Analytics and Computing Laboratory Department of Artificial Intelligence School of Informatics Xiamen University Xiamen361005 China

Many existing space-variant blind image Super-Resolution (SR) methods require the generation of blur kernels for all input pixels. However, the existing methods overlook the possibility of similar blur kernel patterns among adjacent pixels, leading to spatial incoherence and wasteful computational resources. In this study, we introduce an efficient two-stage method for estimating blur kernels. Instead of generating pixel-wise kernels, a limited set of kernels at pixels with distinct information are generated first and the remaining kernels are reconstructed based on this initial set. This novel method, referred to as Anchor Detection and Kernel Reconstruction, comprises two main components: an Anchor Detection Module (ADM) and a Kernel Reconstruction Module (KRM). The objective of ADM is to identify pixels in a given Low-Resolution image that contain rich information, which are called anchors. The corresponding blur kernels, denoted as anchor kernels, are then generated for these identified pixels using a complete backbone network. The remaining blur kernels are reconstructed using KRM with a lightweight interpolation method based on the anchor kernels to enhance spatial consistency among the reconstructed pixels. Extensive experiments demonstrate that the proposed ADKR method maintains comparable performance while estimating only 50% of blur kernels with a full backbone network, reaching approximately 20% reduction in FLOPs. The code has been made available at https://***/xuyimin0926/ADKR. © 2024, The Authors. All rights reserved.

关键词： Pixels

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：