检索结果-南通市图书馆

Evaluation of fiber degree for fish muscle based on the edge feature attention net

Food Bioscience 2022年第3期47卷 635-644页

作者： Junhua Yu Jinlin Zhu Bowen Yan Xidong Jiao Jianlian Huang Jianxin Zhao Hao Zhang Wei Chen Daming Fan State Key Laboratory of Food Science and Technology Jiangnan UniversityWuxi214122China Key Laboratory of Refrigeration and Conditioning Aquatic Products Processing Ministry of Agriculture and Rural AffairsXiamen361022China School of Food Science and Technology Jiangnan UniversityWuxi214122China Fujian Provincial Key Laboratory of Refrigeration and Conditioning Aquatic Products Processing Xiamen361022China Fujian Anjoy Food Share Co.Ltd. Xiamen361022China

The fiber mouthfeel of fish muscle is a highly sought-after goal for surimi gel *** primary aim of research and development has been to quickly and accurately evaluate fiber degree for fish ***,based on the ResNet model,edge feature attentional mechanism was introduced to obtain the edge feature attention net (EFANet) to evaluate fiber degree for fish *** EFANet was trained and tested on a dataset,which was made by collecting microscopic pictures of fish samples with different degrees of *** with the three classic convolutional neural network (CNN) models,the EFANet emphasizes the learning of fiber texture information for fish muscle,reduces the effect of image color change,and significantly improves the detection *** average accuracy and specificity of the EFANet-50 on the testing dataset were 96.22% and 97.92%,respectively,which proved that it can effectively predict the fiber degree of fish muscle.

关键词： Fiber degree Convolutional neural network ResNet attentional mechanism Color change

来源：

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Transformer architecture based on mutual attention for image-anomaly detection

引用

Virtual Reality & Intelligent Hardware 2023年第1期5卷 57-67页

作者： Mengting ZHANG Xiuxia TIAN Department of Computer Science and Techology Shanghai University of Electric PowerShanghai 20090China

Image-anomaly detection, which is widely used in industrial fields. Previous studies that attempted to address this problem often trained convolutional neural network-based models(e.g., autoencoders and generative adversarial networks) to reconstruct covered parts of input images and calculate the difference between the input and reconstructed images. However, convolutional operations are effective at extracting local features, making it difficult to identify larger image anomalies. Method To this end, we propose a transformer architecture based on mutual attention for image-anomaly separation. This architecture can capture long-term dependencies and fuse local and global features to facilitate better image-anomaly detection. Result Our method was extensively evaluated on several benchmarks, and experimental results showed that it improved the detection capability by 3.1% and localization capability by 1.0% compared with state-of-the-art reconstruction-based methods.

关键词： Anomaly detection Swin transformer Feature fusion attentional mechanism Unsupervised learning

来源：

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

An End-to-end Speech Recognition Algorithm based on Attention mechanism

An End-to-end Speech Recognition Algorithm based on Attentio...

引用

第三十九届中国控制会议

作者： Jia-nan Chen Shuang Gao Han-zhe Sun Xiao-hui Liu Zi-ning Wang Yan Zheng College of Information Science and Engineering Northeastern University

End-to-end speech recognition system is a major research field in speech recognition. The most typical model is the end-to-end speech recognition system based on CTC where RNN mining to and time sequence information are adopted, and a series of assumptions of HMM are discarded to obtain a good recognition rate. However, the CTC-based model is more dependent on the speech model and have a longer training cycle. Therefore, in the framework of traditional acoustic model, this paper proposes to train a feature extraction network of spectrogram based on attention mechanism by using prior knowledge. Firstly, it was spliced in the front end based on CTC model, and then the number of layers of cyclic neural network based on CTC model was reduced. Finally, it was combined to retrain. The experimental results show that the training time of the combined model is effectively reduced, and the accuracy of speech recognition is further improved.

关键词： Speech recognition End-to-end technology Endpoint detection attentional mechanism Spectrogram

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Research on Multi-objective Detection Algorithm Based on Improved YOLOv5s Model

Research on Multi-objective Detection Algorithm Based on Imp...

引用

第43届中国控制会议

作者： Chu Tao Jiang Liang Gong Jie Wang Chengchen Xu Fuhui Zhang Dexiang College of Electrical Engineering and Automation Anhui University

Because the real monitoring scene in the city is a complex scene with different situations, there are often various factors such as the change of monitoring object size, local masking, poor contrast and so on. A multi-objective detection algorithm of full scene monitoring based on YOLOv5s model and attention mechanism is proposed. In order to improve the adaptability of the network to the change of object size, a multiscale detection network structure based on YOLOv5s model is designed. Furthermore, an attention mechanism is introduced to focus on more representative parts and suppress less important information. The feature extraction module of attention mechanism is used to suppress background features to enhance object features, and the channel level weight of features is obtained through network learning to improve the network extraction ability of features. At the same time, K-means clustering algorithm is introduced to calculate the initial anchor frame size of the whole scene monitoring data set, so as to improve the convergence speed of the model and the accuracy of object detection. Experimental results demonstrated that the multi-objective detection algorithm based on YOLOv5s model and attention mechanism is effective and efficient in complex urban surveillance scene. Compared with the basic network, the average accuracy of mAP and mAP50 are greatly improved, and the model reasoning time is reduced. In the whole scene monitoring data set, the m AP50 reaches 89.6%. When processing the monitoring video, the Frames Per Second(FPS)is 154 frames/s, which meets the real-time detection requirements of the monitoring site.

关键词： object detection full-scene surveillance improved YOLOv5s model attentional mechanism

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

AGRU-TCDN:Temporal Convolutional Denoising Network for Intelligent Fault Diagnosis

AGRU-TCDN:Temporal Convolutional Denoising Network for Intel...

引用

第32届中国过程控制会议(CPCC2021)

作者： Yongyi Chen Dan Zhang Colleage of Information Engineering Zhejiang University of Technology

Rolling bearing is an important component of electromechanical *** to long-term service in harsh environments such as high speeds and heavy loads,once the failure occurs,the healthy operation of the equipment will be affected at least,and the serious accident will be caused,resulting in huge economic losses and *** order to realize real-time monitoring of the health state of rolling bearings in a high-noise environment,this paper is concerned with the proposition of a new deep learning framework of temporal convolutional denoising network with attention gated recurrent unit(AGRU-TCDN).AGRU-TCDN performs denoising processing on rolling bearing vibration signals through convolutional denoising auto-encoder(CDAE),and then the reconstructed signal is input to the temporal convolutional network(TCN) and attention gated recurrent unit(AGRU),capturing the long-term dependent information of vibration *** effectiveness of this method is verified by experiments with different degrees of noise.

关键词： Temporal Convolutional Network(TCN) Convolutional Denoising Auto-Encoder(CDAE) Gated Recurrent Unit(GRU) attentional mechanism Fault Diagnosis

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

欢迎您,

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

在线全文

在线全文

在线全文

在线全文

在线全文

请选择保存的检索档案：

请选择收藏分类：

通借通还

欢迎您,

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

在线全文

在线全文

在线全文

在线全文

在线全文

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：