检索结果-内蒙古大学图书馆

Attention Guided Global Enhancement and Local Refinement Network for Semantic Segmentation

IEEE TRANSACTIONS ON IMAGE PROCESSING 2022年 31卷 3211-3223页

作者： Li, Jiangyun Zha, Sen Chen, Chen Ding, Meng Zhang, Tianxiang Yu, Hong Univ Sci & Technol Beijing Sch Automat & Elect Engn Beijing 100083 Peoples R China Minist Educ Key Lab Knowledge Automat Ind Proc Beijing 100083 Peoples R China Univ Sci & Technol Beijing Shunde Grad Sch Foshan 528000 Peoples R China Univ Cent Florida Ctr Res Comp Vis Orlando FL 32816 USA Scoop Med Houston TX 77007 USA

The encoder-decoder architecture is widely used as a lightweight semantic segmentation network. However, it struggles with a limited performance compared to a well-designed Dilated-FCN model for two major problems. First, commonly used upsampling methods in the decoder such as interpolation and deconvolution suffer from a local receptive field, unable to encode global contexts. Second, low-level features may bring noises to the network decoder through skip connections for the inadequacy of semantic concepts in early encoder layers. To tackle these challenges, a Global Enhancement Method is proposed to aggregate global information from high-level feature maps and adaptively distribute them to different decoder layers, alleviating the shortage of global contexts in the upsampling process. Besides, aLocal Refinement Module is developed by utilizing the decoder features as the semantic guidance to refine the noisy encoder features before the fusion of these two (the decoder features and the encoder features). Then, the two methods are integrated into a Context Fusion Block, and based on that, a novel Attention guided Global enhancement and Local refinement Network (AGLN) is elaborately designed. Extensive experiments on PASCAL Context, ADE20K, and PASCAL VOC 2012 datasets have demonstrated the effectiveness of the proposed approach. In particular, with a vanilla ResNet-101 backbone, AGLN achieves the state-of-the-art result (56.23% mean IOU) on the PASCAL Context dataset. The code is available at https://***/zhasen1996/AGLN.

关键词： Semantics Decoding Image segmentation Interpolation Convolution Aggregates Context modeling Semantic segmentation encoder-decoder global enhancement local refinement context fusion

来源：评论

学校读者我要写书评

暂无评论

ASCNet: 3D object detection from point cloud based on adaptive spatial context features q

引用

NEUROCOMPUTING 2022年第0期475卷 89-101页

作者： Tong, Guofeng Peng, Hao Shao, Yuyuan Yin, Qijun Li, Zheng Northeastern Univ Coll Informat Sci & Engn Shenyang 110819 Liaoning Peoples R China

This paper presents a novel two-stage 3D point cloud object detector named ASCNet for autonomous driving. Most current works project 3D point clouds to 2D space, whereas the quantization loss in the transformation is inevitable. A Pillar-wise Spatial Context Feature Encoding (PSCFE) module is proposed in the paper to drive the learning of discriminative features and reduce the detailed information loss. The inhomogeneity that existed in 3D object detection from the point clouds, such as the inconsistent number of points in the pillars, the diverse size of Regions of Interest (RoI), should be treated wisely due to the sparsity and the individual specificity. We introduce a length-adaptive RNN-based module to solve the inhomogeneity. A novel backbone combining encoder-decoder and shortcut connection is designed in the paper to learn the multi-scale features for 3D object detection. Additionally, we utilize multiple RoI heads and class-wise NMS to deal with the class imbalance in scenes. Extensive experiments on the KITTI dataset demonstrate that our algorithm achieves competitive performance in 3D bounding box detection and BEV detection. (c) 2021 Elsevier B.V. All rights reserved.

关键词： 3D object detection Point clouds Pillar-wise Inhomogeneity encoder-decoder

来源：评论

学校读者我要写书评

暂无评论

Symbolic Integration by Integrating Learning Models With Different Strengths and Weaknesses

引用

IEEE ACCESS 2022年 10卷 47000-47010页

作者： Kubota, Hazumi Tokuoka, Yuta Yamada, Takahiro G. Funahashi, Akira Keio Univ Grad Sch Fundamental Sci & Technol Ctr Biosci & Informat Yokohama Kanagawa 2238522 Japan Keio Univ Dept Biosci & Informat Yokohama Kanagawa 2238522 Japan

Integration is indispensable, not only in mathematics, but also in a wide range of other fields. A deep learning method has recently been developed and shown to be capable of integrating mathematical functions that could not previously be integrated on a computer. However, that method treats integration as equivalent to natural language translation and does not reflect mathematical information. In this study, we adjusted the learning model to take mathematical information into account and developed a wide range of learning models that learn the order of numerical operations more robustly. In this way, we achieved a 98.80% correct answer rate with symbolic integration, a higher rate than that of any existing method. We judged the correctness of the integration based on whether the derivative of the primitive function was consistent with the integrand. By building an integrated model based on this strategy, we achieved a 99.79% rate of correct answers with symbolic integration. In summary, we have developed a more accurate method of selecting the correct model than the existing method by judging the result of symbolic integration based on whether the output of the model equals the input formula when the output is differentiated.

关键词： Mathematical models Earth Observing System Transformers Solid modeling Deep learning Computational modeling Syntactics Deep learning encoder-decoder supervised learning symbolic integration translation model

来源：评论

学校读者我要写书评

暂无评论

Hand gesture segmentation against complex background based on improved atrous spatial pyramid pooling

引用

JOURNAL OF AMBIENT INTELLIGENCE AND HUMANIZED COMPUTING 2022年第9期14卷 11795-11807页

作者： Cui, Zhenchao Lei, Yu Wang, Yuxiao Yang, Wenzhu Qi, Jing Hebei Univ Sch Cyber Secur & Comp Baoding 071002 Hebei Peoples R China Hebei Univ Hebei Machine Vis Engn Res Ctr Baoding 071002 Hebei Peoples R China

Gesture segmentation is an essential part of gesture detection. The accuracy of gesture detection can be improved by using gesture segmentation to remove the background part un-hand images. However, the inaccurate features of current methods can greatly affect the accuracy of results in segmentation and gesture recognition. In order to solve this problem and obtain accurate features, this paper proposes the improved atrous spatial pyramid pooling (IASPP). IASPP is a pooling layer in convolution neural network, which can refine features by connecting cascade model and parallel model in atrous spatial pyramid pooling. Otherwise, in order to improve the segmentation performance by integrating details and spatial location information at different levels, the IASPP is embedded in the encoder-decoder, and we name the method the improved atrous spatial pyramid pooling-ResNet (IASPP-ResNet) for gesture segmentation. In the experiment part of this paper, we test the proposed method by comparing it with the states of art on the two datasets of OUTHANDS and HGR. It can be seen that IASPP-ResNet can achieve 97.75% Pixel Accuracy and 89.60% MIoU on the OUTHANDS dataset. The Pixel Accuracy and MIoU of the presented method on the HGR dataset can reach 99.09% and 97.52%, respectively. These presented that our method is superior to the states of art.

关键词： Hand gesture segmentation Complex background encoder-decoder Atrous convolution

来源：评论

学校读者我要写书评

暂无评论

An approach to forecasting and filtering noise in dynamic systems using LSTM architectures

引用

NEUROCOMPUTING 2022年第0期500卷 637-648页

作者： Cana, Juan Pedro Llerena Herrero, Jesus Garcia Lopez, Jose Manuel Molina Univ III Madrid Appl Artificial Intelligence Grp GIAA Madrid Spain

Some of the limitations of state-space models are given by the difficulty of modeling certain systems, the filters convergence time, or the impossibility of modeling dependencies in the long term. Having agile and alternative methodologies that allow the modeling of complex problems but still provide solutions to the classic challenges of estimation or filtering, such as the position estimation of a mobile with noisy measurements and unknown motion models, are of high interest. In this work, we address the problem of position estimation of 1-D dynamic systems from a deep learning paradigm, using Long-Short Term Memory (LSTM) architectures designed to solve problems with long term temporal dependencies, in combination with other recurrent networks. A deep neuronal architecture inspired by the encoder decoder language systems is implemented, remarking its limits and finding a solution capable of making predictions of high accuracy with models learnt from training data of a moving object. We use a panel data model for training and validation. In the experimentation, we use sliding overlapping time windows in a recursive and standardized way to avoid the saturation problem of the networks in increasing trend estimates. The results are finally compared with the optimal values from the Kalman filter, obtaining comparable results in error terms. These results show the proposed system has great potential for target tracking.(c) 2022 Elsevier B.V. All rights reserved.

关键词： Deep learning Filtering Forecasting LSTM encoder-decoder Attention

来源：评论

学校读者我要写书评

暂无评论

Research on Image Caption Model Based on Improved Attention Mechanism 9

Research on Image Caption Model Based on Improved Attention ...

引用

9th International Conference on Intelligent Computing and Signal Processing, ICSP 2024

作者： Zhang, Kun'ao Sun, Jinghua School of Computer Science and Technology Xi'an University of Science and Technology Shaanxi Xi'an710054 China

ISBN: (纸本)9798350376548

The basic explanation of the image caption task is that the sentences generated by the model should comprehensively express the content of an image. Existing image caption models face issues such as inadequate utilization of image features and insufficient correlation between extracted image features and semantic information. To address these challenges, a novel image caption model based on asymmetric convolution attention mechanism is proposed. The improved attention mechanism integrates the concept of asymmetric convolution to better utilize the extracted image features. Experimental results on the Flickr8k and Flickr30k image caption datasets demonstrate a significant enhancement in the generated caption sentences based on the asymmetric attention mechanism, as indicated by the improvement in relevant evaluation metrics BLEU2-BLEU4. © 2024 IEEE.

关键词： attention mechanism convolutional neural network encoder-decoder image caption natural language processing

来源：评论

学校读者我要写书评

暂无评论

AI-assisted Silhouettes Generation from Sparse mmWave Sampling

AI-assisted Silhouettes Generation from Sparse mmWave Sampli...

引用

2024 International Conference on Advancements in Power, Communication and Intelligent Systems, APCI 2024

作者： Mohamadi, Fred Tialinx Inc. Irvine United States

ISBN: (纸本)9798350363289

This paper introduces a novel approach leveraging the U-Net algorithm to generate silhouette images of objects using sparse data generated by computers or obtained from mmWave radar. Through this method, exceptional replication of previously unseen images is achieved. The implementation of this algorithm holds significant potential for applications such as collision avoidance on roads or runway incursion. By harnessing sparse data and the power of deep learning, this approach offers promising prospects for enhancing safety measures in various domains. © 2024 IEEE.

关键词： Deep Neural Network encoder-decoder mmWave Object Detection Obscured Environment

来源：评论

学校读者我要写书评

暂无评论

Morphologically Motivated Input Variations and Data Augmentation in Turkish-English Neural Machine Translation

引用

ACM TRANSACTIONS ON ASIAN AND LOW-RESOURCE LANGUAGE INFORMATION PROCESSING 2023年第3期22卷 1-31页

作者： Yirmibesoglu, Zeynep Gungor, Tunga Bogazici Univ Comp Engn Istanbul Turkiye

Success of neural networks in natural language processing has paved the way for neural machine translation (NMT), which rapidly became the mainstream approach in machine translation. Significant improvement in translation performance has been achieved with breakthroughs such as encoder-decoder networks, attention mechanism, and Transformer architecture. However, the necessity of large amounts of parallel data for training an NMT system and rare words in translation corpora are issues yet to be overcome. In this article, we approach NMT of the low-resource Turkish-English language pair. We employ state-of-the-art NMT architectures and data augmentationmethods that exploit monolingual corpora. We point out the importance of input representation for the morphologically rich Turkish language and make a comprehensive analysis of linguistically and non-linguistically motivated input segmentation approaches. We prove the effectiveness of morphologically motivated input segmentation for the Turkish language. Moreover, we show the superiority of the Transformer architecture over attentional encoder-decoder models for the Turkish-English language pair. Among the employed data augmentation approaches, we observe back-translation to be the most effective and confirm the benefit of increasing the amount of parallel data on translation quality. This research demonstrates a comprehensive analysis on NMT architectures with different hyperparameters, data augmentation methods, and input representation techniques, and proposes ways of tackling the low-resource setting of Turkish-English NMT.

关键词： Neural machine translation morphology low-resource Transformer encoder-decoder attention data augmentation word segmentation

来源：评论

学校读者我要写书评

暂无评论

Zero-Shot Learning via Discriminative Dual Semantic Auto-encoder

引用

IEEE ACCESS 2021年 9卷 733-742页

作者： Xing, Nan Liu, Yang Zhu, Hong Wang, Jing Han, Jungong Xian Univ Technol Sch Automat & Informat Engn Xian 710048 Peoples R China Xidian Univ State Key Lab Integrated Serv Networks Xian 710071 Peoples R China Xian Univ Technol Fac Printing Packaging Engn & Digital Media Techn Xian 710048 Peoples R China Aberystwyth Univ Comp Sci Dept Aberystwyth SY23 3FL Dyfed Wales

Zero-shot learning (ZSL) is an effective method to perform the recognition task without any training samples of specific classes. Most existing ZSL models put emphasis on learning an embedding between visual space and semantic space directly. However, few ZSL models research whether the human-designed semantic features are discriminative enough to recognize different classes. Moreover, one-way mapping suffers from the project domain shift problem. In this article, we propose to learn a Discriminative Dual Semantic Auto-encoder (DDSA) based on the encoder-decoder paradigm to solve this problem. DDSA attempts to construct two bidirectional embeddings to connect the visual space and the semantic space with the help of the learned aligned space which includes discriminative information of the visual features and semantic features. Based on the DDSA, we additionally propose a Deep DDSA to capture deep aligned features that are more conducive to zero-shot classification. The key to the proposed framework is that it implicitly exact the principal information from visual space and semantic space to construct aligned features, which is not only semantic-preserving but also discriminative. Extensive experiments on five benchmarks (SUN, CUB, AWA1, AWA2 and aPY) demonstrate the effectiveness of the proposed framework with state-of-the-art performance obtained on both conventional ZSL and generalized ZSL settings.

关键词： Semantics Visualization Training Task analysis Licenses Supervised learning Neural networks Zero-shot learning discriminative encoder-decoder aligned

来源：评论

学校读者我要写书评

暂无评论

A multi-mode traffic flow prediction method with clustering based attention convolution LSTM

引用

APPLIED INTELLIGENCE 2022年第13期52卷 14773-14786页

作者： Huang, Xiaohui Ye, Yuming Wang, Cheng Yang, Xiaofei Xiong, Liyan East China Jiaotong Univ Sch Informat Engn Dept Nanchang 330013 Jiangxi Peoples R China Univ Macau Fac Sci & Technol E11 Macau 999078 Peoples R China

Increasing traffic congestion is a major obstacle to the development of cities. The prediction of traffic flow is very important to city planning and dredging. A good model of flow is able to accurately predict future flow by learning historical flow data. Traffic flow is usually affected by macro and micro factors. At the macro level, the whole city can be divided into different subregions according to the similarity in the traffic flow patterns. At the micro-level, there is a temporal and spatial correlation between the traffic flow of different road sections at di fferent times. In this paper, we propose a multi-mode traffic flow prediction method with Clustering based Attention Convolution LSTM (CACLSTM) to model spatial-temporal data of traffic flow. The framework includes three modules: a convolution LSTM encoding-decoding layer which is used to predict the traffic flow of the next time slice by encoding the historical traffic information, a clustering based attention layer which is able to extract different temporal features by clustering based attention, and an additional factors layer which can integrate weather, wind speed, holidays and other factors to improve the prediction accuracy. The experimental results on Beijing taxis data show that the CACLSTM method performs more effective than the six well-known compared methods.

关键词： Multi-mode encoder-decoder Attention mechanism Traffic flow prediction Spatial-temporal data

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：