检索结果-南通市图书馆

action recognition with Temporal Scale-Invariant Deep Learning Framework

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

China Communications 2017年第2期14卷 163-172页

作者： Huafeng Chen Jun Chen Ruimin Hu Chen Chen Zhongyuan Wang State Key Laboratory of Software Engineering Wuhan University Wuhan 430072 China National Engineering Research Center for Multimedia Software Computer School of Wuhan University Wuhan 430072 China Center for Research in Computer Vision University of Central Florida Orlando FL 32816 USA

Recognizing actions according to video features is an important problem in a wide scope of applications. In this paper, we propose a temporal *** deep learning framework for action recognition, which is robust to the change of action speed. Specifically, a video is firstly split into several *** clips and a keyframe is selected from each *** clip. The spatial and motion features of the keyframe are extracted separately by two Convolutional Neural Networks(CNN) and combined in the convolutional fusion layer for learning the relationship between the features. Then, Long Short Term Memory(LSTM) networks are applied to the fused features to formulate *** temporal clues. Finally, the action prediction scores of the LSTM network are combined by linear weighted summation. Extensive experiments are conducted on two popular and challenging benchmarks, namely, the UCF.101 and the HMDB51 Human actions. On both benchmarks, our framework achieves superior results over the *** methods by 93.7% on UCF.101 and 69.5% on HMDB51, respectively.

关键词： action recognition CNN LSTM

action recognition and Detection Based on Deep Learning: A Comprehensive Summary

在线全文

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2023年第10期77卷 1-23页

作者： Yong Li Qiming Liang Bo Gan Xiaolong Cui College of Information Engineering Engineering University of PAPXi’an710086China PAP of Heilongjiang Province Heihe DetachmentHeihe164300China National Key Laboratory of Science and Technology on Electromagnetic Energy Naval University of EngineeringWuhan430033China Joint Laboratory of Counter Terrorism Command and Information Engineering Engineering University of PAPXi’an710086China

action recognition and detection is an important research topic in computer vision,which can be divided into action recognition and action *** present,the distinction between action recognition and action detection is not clear,and the relevant reviews are not ***,this paper summarized the action recognition and detection methods and datasets based on deep learning to accurately present the research status in this ***,according to the way that temporal and spatial features are extracted from the model,the commonly used models of action recognition are divided into the two stream models,the temporal models,the spatiotemporal models and the transformer models according to the *** this paper briefly analyzes the characteristics of the four models and introduces the accuracy of various algorithms in common data ***,from the perspective of tasks to be completed,action detection is further divided into temporal action detection and spatiotemporal action detection,and commonly used datasets are *** the perspectives of the twostage method and one-stage method,various algorithms of temporal action detection are reviewed,and the various algorithms of spatiotemporal action detection are summarized in ***,the relationship between different parts of action recognition and detection is discussed,the difficulties faced by the current research are summarized in detail,and future development was prospected。

关键词： action recognition action detection deep learning convolutional neural networks dataset

action recognition from Videos with Complex Background via Transfer Learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Donghua University(English Edition) 2010年第2期27卷 199-203页

作者：林贤明李绍滋张洪博刘姝 Department of Cognitive Science Fujian Key Laboratory of the Brain-like Intelligent SystemsXiamen University

Classifier learning methods commonly assume that the training data and the testing data are drawn from the same underlying distribution. However, in many practical situations, this assumption is violated. One example is the practical action videos with complex background and the universal human action databases of Kangliga Tekniska Hogskolan （KTH）. When training data are very scarce, supervised learning is difficult. However, it will cost lots of human and material resources to establish a labeled video set which includes a large amount of videos with complex backgrounds. In this paper, we propose an action recognition framework which uses transfer boosting learning algorithm. By using this algorithm, we can train an action recognition model fitting for most practical situations just relaying on the universal action video dataset and a tiny set of action videos with complex background. And the experiment results show that the performance is improved.

关键词： action recognition transfer adaboost learning maximum mutual information

ResLNet:deep residual LSTM network with longer input for action recognition

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2022年第6期16卷 41-49页

作者： Tian WANG Jiakun LI Huai-Ning WU Ce LI Hichem SNOUSSI Yang WU Institute of Artificial Intelligence Beihang UniversityBeijing100191China School of Automation Science and Electrical Engineering Beihang UniversityBeijing100191China College of Electrical and Information Engineering Lanzhou University of TechnologyLanzhou730050China Institute Charles Delaunay-LM2S FRE CNRS 2019 University of Technology of TroyesTroyes10010France Institute for Research Initiatives Nara Institute of Science and TechnologyNara630-0192Japan

action recognition is an important research topic in video analysis that remains very *** recognition relies on learning a good representation of both spatial information(for appearance)and temporal information(for motion).These two kinds of information are highly correlated but have quite different properties,leading to unsatisfying results of both connecting independent models(e.g.,CNN-LSTM)and direct unbiased co-modeling(e.g.,3DCNN).Besides,a long-lasting tradition on this task with deep learning models is to just use 8 or 16 consecutive frames as input,making it hard to extract discriminative motion *** this work,we propose a novel network structure called ResLNet(Deep Residual LSTM network),which can take longer inputs(e.g.,of 64 frames)and have convolutions collaborate with LSTM more effectively under the residual structure to learn better spatial-temporal representations than ever without the cost of extra computations with the proposed embedded variable stride *** superiority of this proposal and its ablation study are shown on the three most popular benchmark datasets:Kinetics,HMDB51,and *** proposed network could be adopted for various features,such as RGB and optical *** to the limitation of the computation power of our experiment equipment and the real-time requirement,the proposed network is tested on the RGB only and shows great performance.

关键词： action recognition deep learning neural network

Two-Stream Temporal Convolutional Networks for Skeleton-Based Human action recognition

在线全文

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2020年第3期35卷 538-550页

作者： Jin-Gong Jia Yuan-Feng Zhou Xing-Wei Hao Feng Li Christian Desrosiers Cai-Ming Zhang School of Software Shandong UniversityJinan 250101China Department of Software and IT Engineering University of QuebecMontreal H3C 3P8Canada

With the growing popularity of somatosensory interaction devices,human action recognition is becoming attractive in many application ***-based action recognition is effective because the skeleton can represent the position and the structure of key points of the human *** this paper,we leverage spatiotemporal vectors between skeleton sequences as input feature representation of the network,which is more sensitive to changes of the human skeleton compared with representations based on distance and angle *** addition,we redesign residual blocks that have different strides in the depth of the network to improve the processing ability of the temporal convolutional networks(TCNs)for long time dependent *** this work,we propose the two-stream temporal convolutional networks(TSTCNs)that take full advantage of the inter-frame vector feature and the intra-frame vector feature of skeleton sequences in the spatiotemporal *** framework can integrate different feature representations of skeleton sequences so that the two feature representations can make up for each other’s *** fusion loss function is used to supervise the training parameters of the two branch *** on public datasets show that our network achieves superior performance and attains an improvement of 1.2%over the recent GCN-based(BGC-LSTM)method on the NTU RGB+D dataset.

关键词： skeleton action recognition temporal convolutional network(TCN) vector feature representation neural network

action recognition from arbitrary views using 3D-key-pose set

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Electrical and Electronic Engineering in China 2012年第2期7卷 224-241页

作者： Junxia GU Xiaoqing DING Shenjing WANG Department of Electronic Engineering Tsinghua UniversityBeijing 100084China

Recovering three-dimensional （3D） human pose sequence from arbitrary view is very difficult, due to loss of depth information and self-occlusion. In this paper, view-independent 3D-key-pose set is selected from 3D action samples, for the purpose of representing and recognizing those same actions from a single or few cameras without any restriction of the relative orientations between cameras and subjects. First, 3D-key-pose set is selected from the 3D human joint sequences of 3D training action samples that are built from multiple viewpoints. Second, 3D key pose sequence, which matches best with the observation sequence, is selected from the 3D-key- pose set to represent the observation sequence of arbitrary view. 3D key pose sequence contains many discriminative view-independent key poses but cannot accurately describe pose of every frame in the observation sequence. Considering the above reasons, pose and dynamic of action are modeled respectively in this paper. Exemplar-based embedding and probability of unique key pose are applied to model pose property. Complementary dynamic feature is extracted to model these actions that share the same poses but have different dynamic features. Finally, these action models are fused to recognize observation sequence from a single or few cameras. Effectiveness of the proposed approach is demonstrated with experiments on IXMAS dataset.

关键词： Keywords action representation action recognition 3D-key-pose set 3D key pose sequence action models fusion

Arrayed piezoresistive and inertial measurement unit sensor-integrated assistant training tennis racket for multipoint hand pressure monitoring and representative action recognition

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Technological Sciences) 2023年第6期66卷 1746-1756页

作者： YANG MingChuan HOU XiaoJuan WU Hui GUO YangYanHao ZHANG Jie XIE Xin XIAN Shuai WANG Min ZHANG Le QIAN Shuo HE Jian CHOU XiuJian Science and Technology on Electronic Test and Measurement Laboratory North University of ChinaTaiyuan 030051China School of Software North University of ChinaTaiyuan 030051China

Technology-assisted ball training systems have become a research hotspot due to their ability to provide quantitative data for guiding athletes to address their areas of ***,traditional tennis training systems still have some limitations;for instance,they are subjective,expensive,heavy,and *** this research,an assistant training tennis racket,which consists of arrayed flexible sensors and an inertial measurement unit,has been proposed to comprehensively analyze the representative actions’force and *** of MXene as the sensitive material and melamine sponge as the substrate(named MMSS),the flexible sensor exhibited an excellent sensitivity of 5.35 kPa^(-1)(1.1-22.2 kPa)due to the formation of a 3D conductive ***,the sensor retained a high sensitivity of 0.6 k Pa-1in an ultrawide measurement range(22.2-266 kPa).In addition to recognizing the type of hitting action,an artificial intelligence algorithm was introduced to accurately differentiate the five typical motion behaviors with an accuracy rate of 98.2%.This study not only proposes a comprehensive assistant training tennis racket for improving the techniques of tennis enthusiasts but also a new information processing scheme for intelligent sensing and distinction of different movements,which can offer significant application potential in sports big data collection and the Internet of things.

关键词： IMU arrayed piezoresistive sensors multipoint monitoring action recognition intelligent sports

Segment differential aggregation representation and supervised compensation learning of ConvNets for human action recognition

在线全文

学校读者我要写书评

暂无评论

Science China(Technological Sciences) 2024年第1期67卷 197-208页

作者： REN ZiLiang ZHANG QieShi CHENG Qin XU ZhenYu YUAN Shuai LUO DeLin School of Computer Science and Technology Dongguan University of TechnologyDongguan 523808China Shenzhen Institute of Advanced Technology Chinese Academy of SciencesShenzhen 518055China School of Electronic Engineering and Automation Guilin University of Electronic TechnologyGuilin 541004China Department of the mechanical engineering Erlangen-Nuremberg UniversityErlangen 91508Germany School of Aerospace Engineering Xiamen UniversityXiamen 361102China

With more multi-modal data available for visual classification tasks,human action recognition has become an increasingly attractive ***,one of the main challenges is to effectively extract complementary features from different modalities for action *** this work,a novel multimodal supervised learning framework based on convolution neural networks(Conv Nets)is proposed to facilitate extracting the compensation features from different modalities for human action *** on information aggregation mechanism and deep Conv Nets,our recognition framework represents spatial-temporal information from the base modalities by a designed frame difference aggregation spatial-temporal module(FDA-STM),that the networks bridges information from skeleton data through a multimodal supervised compensation block(SCB)to supervise the extraction of compensation *** evaluate the proposed recognition framework on three human action datasets,including NTU RGB+D 60,NTU RGB+D 120,and *** results demonstrate that our model with FDA-STM and SCB achieves the state-of-the-art recognition performance on three benchmark datasets.

关键词： action recognition segment frame difference aggregation supervised compensation learning ConvNets

同方期刊数据库

Attribute-based supervised deep learning model for action recognition

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2017年第2期11卷 219-229页

作者： Kai CHEN Guiguang DING Jungong HAN School of Software Tsinghua University Beijing 100084 China Department of Computer Science Northumbria University Newcastle NEI 8ST UK

Deep learning has been the most popular feature learning method used for a variety of computer vision ap- plications in the past 3 years. Not surprisingly, this tech- nique, especially the convolutional neural networks （Con- vNets） structure, is exploited to identify the human actions, achieving great success. Most algorithms in existence directly adopt the basic ConvNets structure, which works pretty well in the ideal situation, e.g., under stable lighting conditions. However, its performance degrades significantly when the intra-variation in relation to image appearance occurs within the same category. To solve this problem, we propose a new method, integrating the semantically meaningful attributes into deep learning＇s hierarchical structure. Basically, the idea is to add simple yet effective attributes to the category level of ConvNets such that the attribute information is able to drive the learning procedure. The experimental results based on three popular action recognition databases show that the em- bedding of auxiliary multiple attributes into the deep learning framework improves the classification accuracy significantly.

关键词： action recognition convolutional neural network attribute