检索结果-内蒙古大学图书馆

POEM: 1-bit Point-wise Operations based on Expectation-Maximization for Efficient Point Cloud Processing 32

学校读者我要写书评

暂无评论

POEM: 1-bit Point-wise Operations based on Expectation-Maxim...

32nd British Machine Vision Conference, BMVC 2021

作者： Xu, Sheng Li, Yanjing Zhao, Junhe Zhang, Baochang Guo, Guodong Beihang University Beijing China National Engineering Laboratory for Deep Learning Technology and Application Institute of Deep Learning Baidu Research Beijing China

Real-time point cloud processing is fundamental for lots of computer vision tasks, while still challenged by the computational problem on resource-limited edge devices. To address this issue, we implement XNOR-Net-based binary neural networks (BNNs) for an efficient point cloud processing, but its performance is severely suffered due to two main drawbacks, Gaussian-distributed weights and non-learnable scale factor. In this paper, we introduce point-wise operations based on Expectation-Maximization (POEM) into BNNs for efficient point cloud processing. The EM algorithm can efficiently constrain weights for a robust bi-modal distribution. We lead a well-designed reconstruction loss to calculate learnable scale factors to enhance the representation capacity of 1-bit fully-connected (Bi-FC) layers. Extensive experiments demonstrate that our POEM surpasses existing the state-of-the-art binary point cloud networks by a significant margin, up to 6.7%. © 2021. The copyright of this document resides with its authors.

关键词： Computer vision

A human-in-the-loop method for pulmonary nodule detection in CT scans

学校读者我要写书评

暂无评论

Visual Intelligence 2024年第1期2卷 1-13页

作者： Zeng, Qingjie Xie, Yutong Lu, Zilin Xia, Yong National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer Science and Engineering Northwestern Polytechnical University Xi’an 710072 China Australian Institute for Machine Learning The University of Adelaide Adelaide 5000 SA Australia

Automated pulmonary nodule detection using computed tomography scans is vital in the early diagnosis of lung cancer. Although extensive well-performed methods have been proposed for this task, they suffer from the domain shift issue between training and test images. Unsupervised domain adaptation (UDA) methods provide a promising means to mitigate the domain variance;however, their performance is still limited since no target domain supervision is introduced. To make the pulmonary nodule detection algorithm more applicable in clinical practice and further boost the performance across domains, we propose a human-in-the-loop method in a semi-supervised fashion to enhance the model generalization ability when transferred from source domain to target domain. Specifically, we first train a detector model on source domain, and then the pre-trained detector is utilized with our proposed uncertainty-guided sample selection scheme (USSS) to find a few target domain samples worth annotating most and obtain their human annotations. Finally, the annotated and the rest unlabeled target domain samples are used together to refine the pre-trained model via our proposed zoom-in and zoom-out constraint (ZZC) strategy. We evaluate our method on the Nodule Analysis 2016 (LUNA16) and TianChi datasets. Experimental results show that our method surpasses recent competitive methods on source domain and also achieves surprising performance on target domain. © The Author(s) 2024.

关键词： Domain adaptation Human-in-the-loop Pulmonary nodule detection Semi-supervised learning

Feature Selective Transformer for Semantic Image Segmentation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Lin, Fangjian Wu, Tianyi Wu, Sitong Tian, Shengwei Guo, Guodong Institute of Deep Learning Baidu Research Beijing China National Engineering Laboratory for Deep Learning Technology and Application Beijing China

Recently, it has attracted more and more attentions to fuse multi-scale features for semantic image segmentation. Various works were proposed to employ progressive local or global fusion, but the feature fusions are not rich enough for modeling multi-scale context features. In this work, we focus on fusing multi-scale features from Transformer-based backbones for semantic segmentation, and propose a Feature Selective Transformer (FeSeFormer), which aggregates features from all scales (or levels) for each query feature. Specifically, we first propose a Scale-level Feature Selection (SFS) module, which can choose an informative subset from the whole multi-scale feature set for each scale, where those features that are important for the current scale (or level) are selected and the redundant are discarded. Furthermore, we propose a Full-scale Feature Fusion (FFF) module, which can adaptively fuse features of all scales for queries. Based on the proposed SFS and FFF modules, we develop a Feature Selective Transformer (FeSeFormer), and evaluate our FeSeFormer on four challenging semantic segmentation benchmarks, including PASCAL Context, ADE20K, COCO-Stuff 10K, and Cityscapes, outperforming the state-of-the-art. Copyright © 2022, The Authors. All rights reserved.

关键词： Semantic Segmentation

Interaction between internal solitary waves and the seafloor in the deep sea

学校读者我要写书评

暂无评论

deep Underground Science and engineering 2024年第2期3卷 149-162页

作者： Zhuangcai Tian Jinjian Huang Jiaming Xiang Shaotong Zhang Jinran Wu Xiaolei Liu Tingting Luo Jianhua Yue State Key Laboratory of Intelligent Construction and Healthy Operation and Maintenance of Deep Underground Engineering China University of Mining and TechnologyXuzhouChina Research Center for Deep Ocean Science and Underwater Engineering China University of Mining and TechnologyXuzhouChina Frontiers Science Center for Deep Ocean Multispheres and Earth System Key Lab of Submarine Geosciences and Prospecting TechniquesMOECollege of Marine GeosciencesOcean University of ChinaQingdaoChina Institute for Learning Sciences&Teacher Education Australian Catholic UniversityBrisbaneQueenslandAustralia Shandong Provincial Key Laboratory of Marine Environment and Geological Engineering Ocean University of ChinaQingdaoChina Department of Chemical and Biomolecular Engineering National University of SingaporeSingaporeSingapore School of Resources and Geosciences China University of Mining and TechnologyXuzhouChina

Internal solitary wave(ISW),as a typical marine dynamic process in the deep sea,widely exists in oceans and marginal seas *** interaction between ISW and the seafloor mainly occurs in the bottom boundary *** the seabed boundary layer of the deep sea,ISW is the most important dynamic *** study analyzed the current status,hotspots,and frontiers of research on the interaction between ISW and the seafloor by *** on the action of ISW on the seabed,such as transformation and reaction,a large amount of research work and results were systematically analyzed and *** this basis,this study analyzed the wave–wave interaction and interaction between ISW and the bedform or slope of the seabed,which provided a new perspective for an in‐depth understanding of the interaction between ISW and the ***,the latest research results of the bottom boundary layer and marine engineering stability by ISW were introduced,and the unresolved problems in the current research work were *** study provides a valuable reference for further research on the hazards of ISW to marine engineering geology.

关键词： bottom boundary layer interaction internal solitary wave seafloor sediment

Research on deep Clustering Based on Image Data 13th

学校读者我要写书评

暂无评论

Research on Deep Clustering Based on Image Data

13th International Conference on Computer engineering and Networks, CENet 2023

作者： Li, Xuanyu Yang, Houqun Zhang, Xiaoying Yang, Dangui Huang, Jianqiang Gan, Lin College of Computer Science and Technology Hainan University Hainan Haikou570228 China Haikou Key Laboratory of Deep Learning and Big Data Application Technology Hainan University Hainan Haikou570228 China Hainan Haikou China Elmore Family School of Computer and Electrical Engineering Purdue University West Lafayette United States

ISBN: (纸本)9789819992386

Clustering is an important branch of unsupervised tasks, aiming at mining deeper relationships and patterns in data. The quality of feature representation based on image datasets often determines the upper limit of clustering tasks. In recent years, the application of deep learning in deep clustering representation learning module has learned better feature representation, which gradually overcomes the limitation of traditional shallow clustering facing high-dimensional unstructured data. deep clustering has become a popular research direction in the unsupervised field in recent years, and high clustering performance has been obtained in the continuous deepening research. The existing deep clustering research is mainly oriented to various fields of artificial intelligence, including natural language processing (NLP), speech recognition (ASR), computer vision (CV), etc. We take the field of computer vision as the entry point, analyze and study the research progress of deep clustering in computer vision (CV), and deconstruct the deep clustering model into two parts: a detailed classification of the feature extraction module from the perspective of network model architecture, while the data space of clustering is used as the entry point to divide the clustering module, and discuss the deep clustering. Finally, we analyze the datasets and evaluation metrics commonly used in experiments to study deep clustering architectures in computer vision. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2024.

关键词： Feature extraction

Adaptive cross-fusion learning for multi-modal gesture recognition

学校读者我要写书评

暂无评论

Virtual Reality & Intelligent Hardware 2021年第3期3卷 235-247页

作者： Benjia ZHOU Jun WAN Yanyan LIANG Guodong GUO Macao University of Science and Technology Macao 999078China National Laboratory of Pattern Recognition Institute of AutomationChinese Academy of SciencesBeijing 100190China Baidu Research Beijing 100193Chinaand National Engineering Laboratory for Deep Learning Technology and ApplicationBeijing 100193China

Background Gesture recognition has attracted significant attention because of its wide range of potential *** multi-modal gesture recognition has made significant progress in recent years,a popular method still is simply fusing prediction scores at the end of each branch,which often ignores complementary features among different modalities in the early stage and does not fuse the complementary features into a more discriminative *** This paper proposes an Adaptive Cross-modal Weighting(ACmW)scheme to exploit complementarity features from RGB-D data in this *** scheme learns relations among different modalities by combining the features of different data *** proposed ACmW module contains two key functions:(1)fusing complementary features from multiple streams through an adaptive one-dimensional convolution;and(2)modeling the correlation of multi-stream complementary features in the time *** the effective combination of these two functional modules,the proposed ACmW can automatically analyze the relationship between the complementary features from different streams,and can fuse them in the spatial and temporal *** Extensive experiments validate the effectiveness of the proposed method,and show that our method outperforms state-of-the-art methods on IsoGD and NVGesture.

关键词： Gesture recognition Multi-modal fusion RGB-D

CATrans: Context and Affinity Transformer for Few-Shot Segmentation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Shan Wu, Tianyi Wu, Sitong Guo, Guodong Australian National University Canberra Australia Institute of Deep Learning Baidu Research Beijing China National Engineering Laboratory for Deep Learning Technology and Application Beijing China

Few-shot segmentation (FSS) aims to segment novel categories given scarce annotated support images. The crux of FSS is how to aggregate dense correlations between support and query images for query segmentation while being robust to the large variations in appearance and context. To this end, previous Transformer-based methods explore global consensus either on context similarity or affinity map between support-query pairs. In this work, we effectively integrate the context and affinity information via the proposed novel Context and Affinity Transformer (CATrans) in a hierarchical architecture. Specifically, the Relation-guided Context Transformer (RCT) propagates context information from support to query images conditioned on more informative support features. Based on the observation that a huge feature distinction between support and query pairs brings barriers for context knowledge transfer, the Relation-guided Affinity Transformer (RAT) measures attention-aware affinity as auxiliary information for FSS, in which the self-affinity is responsible for more reliable cross-affinity. We conduct experiments to demonstrate the effectiveness of the proposed model, outperforming the state-of-the-art methods. © 2022, CC BY.

关键词： Image segmentation

FM-ViT: Flexible Modal Vision Transformers for Face Anti-Spoofing

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liu, Ajian Tan, Zichang Yu, Zitong Zhao, Chenxu Wan, Jun Liang, Yanyan Lei, Zhen Zhang, Du Li, Stan Z. Guo, Guodong Beijing China The Institute of Deep Learning Baidu Research National Engineering Laboratory for Deep Learning Technology and Application China The Great Bay University Dongguan523000 China The Mininglamp Academy of Sciences Mininglamp Technology China The Macau University of Science and Technology China The Westlake University China

The availability of handy multi-modal (i.e., RGB-D) sensors has brought about a surge of face anti-spoofing research. However, the current multi-modal face presentation attack detection (PAD) has two defects: (1) The framework based on multi-modal fusion requires providing modalities consistent with the training input, which seriously limits the deployment scenario. (2) The performance of ConvNet-based model on high fidelity datasets is increasingly limited. In this work, we present a pure transformer-based framework, dubbed the Flexible Modal Vision Transformer (FM-ViT), for face anti-spoofing to flexibly target any single-modal (i.e., RGB) attack scenarios with the help of available multi-modal data. Specifically, FM-ViT retains a specific branch for each modality to capture different modal information and introduces the Cross-Modal Transformer Block (CMTB), which consists of two cascaded attentions named Multi-headed Mutual-Attention (MMA) and Fusion-Attention (MFA) to guide each modal branch to mine potential features from informative patch tokens, and to learn modality-agnostic liveness features by enriching the modal information of own CLS token, respectively. Experiments demonstrate that the single model trained based on FM-ViT can not only flexibly evaluate different modal samples, but also outperforms existing single-modal frameworks by a large margin, and approaches the multi-modal frameworks introduced with smaller FLOPs and model parameters. Copyright © 2023, The Authors. All rights reserved.

关键词： Modal analysis

Dynamic Group Transformer: A General Vision Transformer Backbone with Dynamic Group Attention

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Liu, Kai Wu, Tianyi Liu, Cong Guo, Guodong Sun Yat-Sen University Guangzhou China Institute of Deep Learning Baidu Research Beijing China National Engineering Laboratory for Deep Learning Technology and Application Beijing China

Recently, Transformers have shown promising performance in various vision tasks. To reduce the quadratic computation complexity caused by each query attending to all keys/values, various methods have constrained the range of attention within local regions, where each query only attends to keys/values within a hand-crafted window. However, these hand-crafted window partition mechanisms are data-agnostic and ignore their input content, so it is likely that one query maybe attends to irrelevant keys/values. To address this issue, we propose a Dynamic Group Attention (DG-Attention), which dynamically divides all queries into multiple groups and selects the most relevant keys/values for each group. Our DG-Attention can flexibly model more relevant dependencies without any spatial constraint that is used in hand-crafted window based attention. Built on the DG-Attention, we develop a general vision transformer backbone named Dynamic Group Transformer (DGT). Extensive experiments show that our models can outperform the state-of-the-art methods on multiple common vision tasks, including image classification, semantic segmentation, object detection, and instance segmentation. Copyright © 2022, The Authors. All rights reserved.

关键词：