检索结果-内蒙古大学图书馆

Piktor-O-Bot: Integrated image processing algorithms for portrait drawing robot applications

Measurement: Sensors 2025年 38卷

作者： Hoang, Anh Tuan Csempesz, János Cserteg, Tamás Viharos, Zsolt János Budapest Hungary Faculty of Economics and Business John von Neumann University Kecskemét Hungary Doctoral School of Informatics ELTE Eötvös Loránd University BudapestH-1117 Hungary

Collaborative robots, or cobots, have emerged as vital assets in numerous industries, reshaping operational paradigms and fostering human-machine collaboration. This paper introduces Piktor-o-bot, a Universal Robots 5 series portrait drawing robot, which integrates advanced image processing algorithms with collaborative robotic functionalities to streamline and enhance portrait creation. By utilizing neural networks for human face detection, background removal, edge detection, semantic segmentation, clothes segmentation, and eye detection, Piktor-o-bot aims to achieve lifelike depictions efficiently. The proposed system integrates these specialized neural networks, innovative algorithms, and collaborative robotic capabilities to offer a promising and optimized solution. The individual components and their interactions contribute to the overall success of the Pictor-o-bot system, as demonstrated by its performance at several exhibitions. © 2024

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

A Survey on Test Input Selection and Prioritization for Deep neural networks

A Survey on Test Input Selection and Prioritization for Deep...

引用

International Symposium on System and Software Reliability (ISSSR)

作者： Shengrong Wang Dongcheng Li Hui Li Man Zhao W. Eric Wong School of Computer Science China University of Geosciences Wuhan China Department of Computer Science California State Polytechnic University - Humboldt Arcata USA Department of Computer Science University of Texas at Dallas Richardson USA

ISBN: (数字)9798350362930

ISBN: (纸本)9798350362947

With the breakthrough advancements of deep neural network technology in applications such as image processing, autonomous driving, and speech recognition, the testing of deep neural network models becomes crucial to ensure their performance and reliability. The selection of test inputs is a critical step in the testing process. Prioritizing test inputs and selecting the most impactful ones can help improve testing efficiency to identify potential problems and deficiencies in the model as early as possible, given the large size of the test dataset and the high cost of annotation. In order to gain a deeper understanding of the research progress in the field of test input selection for deep neural networks, this paper conducts a survey of academic papers in the field over the past years. A systematic review of existing research outcomes is presented, focusing on the criteria, methods, and priority ranking for the selection of test inputs in deep neural networks. Additionally, the paper offers insights into future challenges for test input selection in deep neural networks.

关键词： Surveys Uncertainty Systematics Reviews artificial neural networks Speech recognition Robustness

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Models Selection and Adaptation for a Typical Problem of X-Ray image processing

Deep Learning Models Selection and Adaptation for a Typical ...

引用

2021 Innovations in Intelligent Systems and applications Conference, ASYU 2021

作者： Sheshulin, Kirill Chuvakov, Aleksandr Ivaschenko, Anton Samara State Technical University Computer Science Department Samara Russia

ISBN: (纸本)9781665434058

The paper explores a problem of artificial neural networks comparative analysis and adaptation while solving a typical task of image analysis and pattern recognition. This article examines four neural networks using different optimizers to solve a problem of pneumonia detection on X-ray images. The study implements convolutional neural networks, as well as the method of their transfer learning. The architecture of neural networks and the results of their work with various metrics that assess the success of this model are presented. The problem of interpretability of the decision made by a neural network can be solved using a special visualization method presented in the study. Advantages of the proposed research are related with a possibility to use the convolutional neural networks in practical applications. © 2021 IEEE.

关键词： Intelligent systems

来源：评论

学校读者我要写书评

暂无评论

A genetic programming-based convolutional deep learning algorithm for identifying COVID-19 cases via X-ray images

引用

artificial INTELLIGENCE IN MEDICINE 2023年第1期142卷 102571-102571页

作者： Najaran, Mohammad Hassan Tayarani Univ Hertfordshire Sch Comp Sci Hatfield England

Evolutionary algorithms have been successfully employed to find the best structure for many learning algorithms including neural networks. Due to their flexibility and promising results, Convolutional neural networks (CNNs) have found their application in many image processing applications. The structure of CNNs greatly affects the performance of these algorithms both in terms of accuracy and computational cost, thus, finding the best architecture for these networks is a crucial task before they are employed. In this paper, we develop a genetic programming approach for the optimization of CNN structure in diagnosing COVID-19 cases via X-ray images. A graph representation for CNN architecture is proposed and evolutionary operators including crossover and mutation are specifically designed for the proposed representation. The proposed architecture of CNNs is defined by two sets of parameters, one is the skeleton which determines the arrangement of the convolutional and pooling operators and their connections and one is the numerical parameters of the operators which determine the properties of these operators like filter size and kernel size. The proposed algorithm in this paper optimizes the skeleton and the numerical parameters of the CNN architectures in a co-evolutionary scheme. The proposed algorithm is used to identify covid-19 cases via X-ray images.

关键词： Genetic programming Deep learning Optimization Evolutionary algorithms COVID-19 Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Bearing Fault image Recognition in Strong Noise Based on Optimized CNN Architecture

Bearing Fault Image Recognition in Strong Noise Based on Opt...

引用

image processing and Computer applications (ICIPCA), IEEE International Conference on

作者： Qiang Li Haowen Gai Kaijie Hu Northeastern University Shenyang China Tianjin College University of Science and Technology Beijing Tianjin China Zhonghuan Information College Tianjin University of Technology Tianjin China

ISBN: (数字)9798350360240

ISBN: (纸本)9798350384161

To address the problem of low accuracy and poor stability of bearing diagnostic models under strong background noise, a bearing fault image recognition method is proposed that reduces the randomness of the model by avoiding the need for artificial parameterization, which can introduce random factors. This method is based on hyperparameter optimization of the GoogLeNet convolutional neural network model and decision fusion. First of all, two-dimensional wavelet time-frequency variation of the original signal of the bearing vibration to construct an image dataset, one-dimensional classification is into a two-dimensional image problem; Secondly, three lightweight convolutional neural network architectures are selected for the noise immunity test, to get the best noise-resistant network architecture GoogLeNet; Finally, hyper-parameter optimization is performed for the network, comparing the mesh method, the progressive mesh method, and the group optimization algorithm, respectively, and getting The optimal parameters are then analyzed for decision fusion and visualization of the network. To verify the method proposed in this paper, we used the open Case Western Reserve University bearing dataset. The experimental verification demonstrates that the proposed method achieves a correctness rate of 94.33% in decision fusion under noise, with good accuracy and stability.

关键词： Accuracy image recognition Computer architecture Power system stability White noise Stability analysis Robustness Classification algorithms Convolutional neural networks Optimization

来源：评论

学校读者我要写书评

暂无评论

Fuzzy Systems and Data Mining vii - Proceedings of FSDM 2021

Fuzzy Systems and Data Mining VII - Proceedings of FSDM 2021

引用

7th International Conference on Fuzzy Systems and Data Mining, FSDM 2021

ISBN: (纸本)9781643682143

The proceedings contain 52 papers. The topics discussed include: how artificial intelligence may impact your job;a deep neural network model for the prediction of major adverse cardiovascular event occurrences in patients with non-ST-elevation myocardial infarction;do reviews influence real estate marketing: the experience combing with natural language processing;ranking of trapezoidal bipolar fuzzy numbers based on a new improved score function;interpretable dual-feature recommender system using reviews;facial expression recognition and image description generation in Vietnamese;hierarchical digital control system performance;and generalized-multiquadric radial basis function neural networks (RBFNs) with variable shape parameters for function recovery.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Apply Yolov4-Tiny on an FPGA-Based Accelerator of Convolutional neural Network for Object Detection 2

Apply Yolov4-Tiny on an FPGA-Based Accelerator of Convolutio...

引用

2022 2nd International Conference on artificial Intelligence and Industrial Technology applications, AIITA 2022

作者： Zhang, Fengxi Li, Yuying Ye, Zhihao School of Telecommunications Engineering Xidian University No.266 Xinglong Section of Xi Feng Rd Shaanxi Province Xi'an710126 China School of Artificial Intelligent Xidian University No.266 Xinglong Section of Xi Feng Rd Shaanxi Province Xi'an710126 China

With the continuous expansion of neural Network technology in the artificial intelligence field, for example, image recognition and retrieval, object detection, pixel processing, automatic speech generation, etc., Convolutional neural networks (CNN) and Deep Learning of neural networks (DNN) have made apparent breakthroughs. To improve the inference speed of images, the combination of FPGA-based acceleration and multiple model quantization methods has become one of the most contemporary alternative methods. This paper designed an FPGA-Based acceleration scheme combining software and hardware and effectively applied it to the Yolov4-Tiny object detection model, realizing the accelerated detection inference process from the original 6-7mins to 383ms. First, it chose the static quantization method of fixed-point numbers, fixed the position of the decimal point, and then added Batch Norm between the convolutional layer and the activation function to form a connection structure. Second, it further improved inference speed on an FPGA with a version of ZYNQ-7020 by increasing the bandwidth cap and reducing bandwidth requirements, employing a massive pipeline design. Finally, in the test of the Coco dataset, the plan has completed a substantial acceleration of the average inference speed of the Yolov4-Tiny object detection model from 7.13mins/Picture to 498.89ms/Picture, which has a high application value in the field of object detection. It dramatically improves the inference speed as well as keeps the average accuracy above 0.95. © Published under licence by IOP Publishing Ltd.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Tool wear and remaining useful life prediction in micro-milling along complex tool paths using neural networks

引用

JOURNAL OF MANUFACTURING PROCESSES 2021年 71卷 679-698页

作者： Bagri, Sumant Manwar, Ashish Varghese, Alwin Mujumdar, Soham Joshi, Suhas S. Indian Inst Technol Dept Mech Engn Mumbai 400076 Maharashtra India

Monitoring tool wear state and prediction of the remaining useful life in micro-milling help avoid down time due to unalarmed failure of the cutting tool. In this regard, investigations based on experiments, and data-based models, along straight tool paths have been reported in the literature. However, industrial applications often require machining along complex tool paths, where the tool failure possibilities increase multifold, have not been studied elaborately in the literature. Therefore, the main objective of this work is to develop a data-based model, which can predict the tool wear state, and remaining useful life of the tool, considering the effect of tool path complexities due to varying tool path radius in micro-milling. The data for developing the model were generated by performing micro-milling experiments under different processing parameters, such as feed, depth of cut, spindle rotation speed, and tool path radius. The tool images were captured in-situ without removing the tool from the spindle, and image binarization and alignment operations were performed to extract corresponding cutting edge wear features, such as diameter reduction as well as the wear of individual cutting edges. The tool wear classification criteria were defined to categorize the tool condition in three regions: initial wear, steadystate wear, and critical wear. To capture and model the complex mechanism of micro tool wear and failure, artificial neural networks, as well as deep belief networks, were implemented to predict the wear state as well as the remaining-useful-life (RUL) of the tools. It was found that the wear rate increased with increasing tool path radii and the larger radii could lead to catastrophic tool failure. The neural-network models gave 93-99% accuracy for the prediction of wear state classification and RUL.

关键词： Micro end-milling Tool path radius Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Untargeted Backdoor Attack Against Object Detection 48

Untargeted Backdoor Attack Against Object Detection

引用

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

作者： Luo, Chengxiao Li, Yiming Jiang, Yong Xia, Shu-Tao Tsinghua University Tsinghua Shenzhen International Graduate School China Research Center of Artificial Intelligence Peng Cheng Laboratory China

ISBN: (纸本)9781728163277

Recent studies revealed that deep neural networks (DNNs) are exposed to backdoor threats when training with third-party resources (such as training samples or backbones). The back-doored model has promising performance in predicting benign samples, whereas its predictions can be maliciously manipulated by adversaries based on activating its backdoors with pre-defined trigger patterns. Currently, most of the existing backdoor attacks were conducted on the image classification under the targeted manner. In this paper, we reveal that these threats could also happen in object detection, posing threatening risks to many mission-critical applications (e.g., pedestrian detection and intelligent surveillance systems). Specifically, we design a simple yet effective poison-only backdoor attack in an untargeted manner, based on task characteristics. We show that, once the backdoor is embedded into the target model by our attack, it can trick the model to lose detection of any object stamped with our trigger patterns. We conduct extensive experiments on the benchmark dataset, showing its effectiveness in both digital and physical-world settings and its resistance to potential defenses. © 2023 IEEE.

关键词： AI Security Backdoor Attack Object Detection Physical Attack Trustworthy ML

来源：评论

学校读者我要写书评

暂无评论

A limited-size ensemble of homogeneous CNN/LSTMs for high-performance word classification

引用

neural COMPUTING & applications 2021年第14期33卷 8615-8634页

作者： Ameryan, Mahya Schomaker, Lambert Univ Groningen Artificial Intelligence & Cognit Engn Fac Sci & Engn Groningen Netherlands

The strength of long short-term memory neural networks (LSTMs) that have been applied is more located in handling sequences of variable length than in handling geometric variability of the image patterns. In this paper, an end-to-end convolutional LSTM neural network is used to handle both geometric variation and sequence variability. The best results for LSTMs are often based on large-scale training of an ensemble of network instances. We show that high performances can be reached on a common benchmark set by using proper data augmentation for just five such networks using a proper coding scheme and a proper voting scheme. The networks have similar architectures (convolutional neural network (CNN): five layers, bidirectional LSTM (BiLSTM): three layers followed by a connectionist temporal classification (CTC) processing step). The approach assumes differently scaled input images and different feature map sizes. Three datasets are used: the standard benchmark RIMES dataset (French);a historical handwritten dataset KdK (Dutch);the standard benchmark George Washington (GW) dataset (English). Final performance obtained for the word-recognition test of RIMES was 96.6%, a clear improvement over other state-of-the-art approaches which did not use a pre-trained network. On the KdK and GW datasets, our approach also shows good results. The proposed approach is deployed in the Monk search engine for historical-handwriting collections.

关键词： Coding scheme Ensemble system End-to-end convolutional long short-term memory Connectionist temporal classification

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：