检索结果-内蒙古大学图书馆

19th IEEE International Conference on Mechatronics and Automation (IEEE ICMA)

作者： Ma, Zhongli Zeng, Yuehan Zhang, Linshuai Li, Jiadi Chengdu Univ Informat Technol Dept Control Engn CULT Chengdu Sichuan Peoples R China

ISBN: (数字)9781665408530

ISBN: (纸本)9781665408530;9781665408523

In order to solve the problem of high error rate and poor real-time performance in the workpieces sorting process for traditional industrial robotic arms, this paper designed a vision robotic arm testing platform with real-time processing ability, and proposes a kind of workpieces sorting method based on improved YOLOv5 used to the vision robotic arm. By replacing the focus layer in the YOLOv5 backbone network, embedding the coordinate attention module, which re-weights the feature maps from the channel and spatial, improves the object detection accuracy of the YOLOv5 model. The workpiece sorting test platform consists of an NVIDIA Jetson nano controller and a vision robotic arm. The hand-eye calibration of the robotic arm is completed by the Zhang Zhengyou calibration method and the TsarLenz method. The workpiece target image was collected, tagged and data augmented to create the target workpiece dataset. And use TensorRT to optimize the inference acceleration of the model to adapt to the hardware platform requirements. The test shows that the improved YOLOv5 model can well ensure the stable operation of the test platform, and improve the accuracy and real -time performance of workpiece target recognition.

关键词： deep learning Robotic arm sorting YOLO v5 Object detection Attention mechanism TensorRT

来源：评论

学校读者我要写书评

暂无评论

deep view synthesis with compact and adaptive Multiplane images

引用

SIGNAL processing-image COMMUNICATION 2022年 107卷

作者： Navarro, Julia Sabater, Neus Univ Illes Balears DMI IAC3 Palma De Mallorca Spain InterDigital Cesson-sevigne France

Multiplane images (MPIs) have shown to be excellent scene representations to synthesize new scene views. Indeed, MPIs are able to model challenging occlusions and reflections, and allow to render novel images in real time and with angular consistency. However, their memory footprint constitutes their major limitation. In this work, we propose a learning-based method that computes compact and adaptive MPIs. Our network promotes sparsity in the MPIs to only keep the necessary scene information. Besides, we adapt the depth sampling to the given scene to optimize the available memory and increase the synthesis quality with a restricted number of planes. Moreover, in contrast to recent work, our approach does not need individual training per scene and is able to generalize well to unseen scenarios. An extensive evaluation shows the superiority of our approach with respect to the state of the art on diverse view synthesis datasets.

关键词： View synthesis Multiplane image deep learning

来源：评论

学校读者我要写书评

暂无评论

Neural Architecture Search for real-time Driver Behavior Recognition 4

Neural Architecture Search for Real-Time Driver Behavior Rec...

引用

4th International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2022

作者： Seong, Jaeho Lee, Chaehyun Han, Dong Seog Kyungpook National University Department of Future Automotive and IT Convergence Daegu Korea Republic of Kyungpook National University School of Electronic and Electrical Engineering Daegu Korea Republic of

ISBN: (纸本)9781665458184

Driver behavior recognition (DBR) helps to ensure driver safety by alerting drivers about potential hazards and minimizing them. In this paper, we use deep learning-based neural architecture search (NAS) to classify driver behavior. In the NAS method, a reinforcement learning algorithm is used, and the neural network architecture is quickly searched by sharing the weights of the parameters. Most DBR models focus on accuracy, while high processing speed is required in order to be applied to actual vehicles. In addition, since the driver monitoring system (DMS) includes complex algorithms based on deep learning, it requires a DBR model that takes this into account. We collect our own data set for driver behavior classification and recognize four common driving behaviors: general driving, mobile phone use, food intake, and smoking. The proposed model on our own data set collected through experiments has better performance and lower network cost than the previous lightweight classification model. © 2022 IEEE.

关键词： deep learning

来源：评论

学校读者我要写书评

暂无评论

MAKE: A Combined Autoencoder to Detect Adversarial Examples 6

MAKE: A Combined Autoencoder to Detect Adversarial Examples

引用

6th International Conference on Signal and image processing, ICSIP 2021

作者： He, Zhaoxiang Yu, Zihan Chen, Liquan Qin, Zhongyuan Zhang, Qunfang Zhang, Yipeng Southeast University Key Laboratory of Computer Networking Technology of Jiangsu Province School of Cyber Science and Engineering Nanjing China Air-Defence Institute Nanjing Campus Nanjing China

ISBN: (纸本)9780738133737

With the continuous development and maturity of deep learning technologies, security issues in deep learning are also getting more and more attention. The generation of adversarial examples makes scholars more aware of this point. The addition of small disturbance in the original image can cause the misclassification of images by deep learning models, which seriously hinders the development and popularization of deep learning technology in the future. Therefore, a method based on MSE and KL AutoEncoder (MKAE) to detect adversarial examples is proposed. By using MSE and KL divergence together, it is proved that MKAE can resist various types of adversarial attacks. At the same time, compared with the existing feature squeezing and MagNet detection algorithms, the detection accuracy is improved. This method is not dependent on the specific attack mode, which is a movable detection and defense model. © 2021 IEEE.

关键词： deep learning

来源：评论

学校读者我要写书评

暂无评论

Yoga Pose Recognition with real time Correction using deep learning

Yoga Pose Recognition with Real time Correction using Deep L...

引用

Sustainable Computing and Data Communication Systems (ICSCDS), International Conference on

作者： Vinay Chethan Reddy Pala Sreekar Kamatagi Shyamsunder Jangiti K Swaraja K Reddy Madhavi Gs Naveen Kumar ECE GRIET Hyderabad India School of Computing Mohan Babu University Tirupati India CSE Malla Reddy University Hyderabad India

In day-to-day life, it can be difficult for a person to devote his time to attend Yoga classes. In Yoga sessions, there might be a lack of individual attention for each person. While performing poses, incorrect muscle usage might lead to long-term muscle pain, back pain or many other deformities. To solve the aforementioned problems, a web application is built where a person can correct yoga pose. The Proposed methodology is working with TensorFlow lite Pose detection python module for recognizing human action based on Yoga Pose Classification using image processing and deep learning. The Objective of pose estimation is for monitoring the movement of human pose for distinct exercises. From this, the recognition of yoga poses can be done using backend part and wrongly recognized yoga poses can be corrected using frontend part. A real-time test is also carried out within a group of 5 people (three men and two women), and the accuracy attained is around 90%. Using deep learning, the proposed model accuracy is evaluated by fitting the training data and predicting it over the testing data which is estimated to be around 98%.

关键词： deep learning Training Pain Computational modeling Training data Muscles Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Pothole Detection Using deep Convolutional Neural Network

Pothole Detection Using Deep Convolutional Neural Network

引用

Future of Information and Communication Conference, FICC 2021

作者： Prakash, Ved Velampalli, Sirisha University of Hyderabad Hyderabad India CR Rao AIMSCS University of Hyderabad Campus Hyderabad India

ISBN: (纸本)9783030731021

In this work, we explored the possibility of developing a deep learning model which can detect potholes on roads in real time with maximum accuracy and minimum inference delay. We compared the results of image processing-based methods with deep learning methods and we found improved results using deep learning methods. We experimented with changing depth of convolutional layers and their effect on confusion matrix and accuracy. We will also see the results of pothole segmentation and how it localises the result and the improvement in overall accuracy. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Deformable alignment of longitudinal postoperative brain GBM scans using deep learning

Deformable alignment of longitudinal postoperative brain GBM...

引用

Medical Imaging Conference - image processing

作者： Lao, Yi Yu, Victoria Chang, Eric Yang, Wensha Sheng, Ke Univ Calif Los Angeles Dept Radiat Oncol Los Angeles CA 90024 USA Univ Southern Calif Dept Radiat Oncol Keck Sch Med Los Angeles CA 90007 USA

ISBN: (纸本)9781510633940

Longitudinal brain alignment is critical for disease monitoring and adaptive treatment planning in glioblastoma (GBM) patients. However, the current methods are either non-adaptive to pathological brains, or time and labor-intensive. Here, we aim to develop a novel deep-learning-based framework for longitudinal postoperative brain GBM scan registration. The proposed pathology adaptive registration framework (PARF) adopts a double UNET architecture: a 2D 7-level UNET, NETseg, for pathology segmentation, and a 3D 5-level UNET, NETseg, for unsupervised image registration, connected through a spatial transformer and a volume combiner. NETseg was first trained separately and then combined with NETseg for pathology adaptive registration training. In aggregated registration testing of PARF, 36 registrations from 18 intra-subject pairs of post-operative follow-up MR scans were selected, and the results were compared to those from current state-of-the-art methods as well as non-adaptive NETseg alone. PARF is significantly faster and more accurate than comparison methods, in terms of sum-of-squared differences, segmentation alignment dice coefficients, and landmark mislignment errors. PARF may pave the path for various clinical and research applications that depend on the accurate registration of GBM longitudinal images.

关键词： Brain registration Unsupervised deep learning Brain GBM deep-learning-based registration

来源：评论

学校读者我要写书评

暂无评论

real-time Ultrasound image Despeckling Using Mixed-Attention Mechanism Based Residual UNet

引用

IEEE ACCESS 2020年 8卷 195327-195340页

作者： Lan, Yancheng Zhang, Xuming Huazhong Univ Sci & Technol Sch Life Sci & Technol Minist Educ Key Lab Mol Biophys Wuhan 430074 Peoples R China

Ultrasound imaging has been widely used for clinical diagnosis. However, the inherent speckle noise will degrade the quality of ultrasound images. Existing despeckling methods cannot deliver sufficient speckle reduction and preserve image details well at high noise corruption and they cannot realize real-time ultrasound image denoising. With the popularity of deep learning, supervised learning for image denoising has recently attracted considerable attention. In this paper, we have proposed a novel residual UNet using mixed-attention mechanism (MARU) for real-time ultrasound image despeckling. In view of the signal-dependent characteristics of speckle noise, we have designed an encoder-decoder network to reconstruct the despeckled image by extracting features from the noisy image. Furthermore, a lightweight mixed-attention block is proposed to effectively enhance the image features and suppress some speckle noise during the encoding phase by using separation and re-fusion strategy for channel and spatial attention. Besides, we have graded the speckle noise levels with a certain interval and designed an algorithm to estimate the noise levels for despeckling real ultrasound images. Experiments have been done on the natural images, the synthetic image, the image simulated using Field II and the real ultrasound images. Compared with existing despeckling methods, the proposed network has achieved the state-of-the-art despeckling performance in terms of subjective human vision and such quantitative indexes as peak signal to noise ratio (PSNR), structural similarity (SSIM), equivalent number of looks (ENL) and contrast-to-noise ratio (CNR).

关键词： Ultrasonic imaging Speckle real-time systems image denoising Noise reduction deep learning Feature extraction Ultrasound image speckle noise supervised learning mixed-attention mechanism residual UNet

来源：评论

学校读者我要写书评

暂无评论

real time Object Detection in Video Surveillance Using Fast-D Algorithm

Real Time Object Detection in Video Surveillance Using Fast-...

引用

2023 IEEE International Conference on Research Methodologies in Knowledge Management, Artificial Intelligence and Telecommunication Engineering, RMKMATE 2023

作者： Madhan, K. Shanmugapriya, N. Department of Computer Science and Engineering Dhanalakshmi Srinivasan University Trichy621 112 India

ISBN: (纸本)9798350305708

With the proliferation of video surveillance devices, the value of computer-assisted detection of anomalous occurrences in video streams has increased. Abnormal prevalence can also be viewed as an abnormal dip compared to the normal course. However, because the relationship between normal and abnormal is particularly unbalanced with respect to reality, many unnatural occurrences have become less frequent. Techniques have been proposed to detect unnatural video events based on both Convolutional Neural Networks (CNNs) and instance-based communication. This strategy has been used previously to recognize the need to localize anomalous video events within pixel-level regions. First, we use a Gaussian background model to accurately identify moving objects in the movie, and then use image processing techniques to capture the relevant regions of the identified moving objects. Finally, according to the purpose of use, the prepared according to the suction function from the combined area will be distributed among the systems, and then used according to the development of some core field packages. Finally, the multi-instance learning model learns how to use the normalized Embark-Head quotes approach to make pixel-level predictions. However, the (Fast-D) target detection method is applied depending on the detection of two-lane accidents. Based on our experimental results, video exception detection methods based on CNN or sparse illustration commands can accurately detect strange occurrences in pixel-composed environments. With this in mind, the reason I created this essay challenge was to find the first solution to the same problem using sound teaching techniques. This was done to avoid the need to include ethnic sources within the scope of anomalous activity to the extent that one would expect live speech to be observed outside of a system of rules. © 2023 IEEE.

关键词： Abnormal Detection CNN deep learning Fast-D Object Detection

来源：评论

学校读者我要写书评

暂无评论

Sounding out the hidden data: A concise review of deep learning in photoacoustic imaging

引用

EXPERIMENTAL BIOLOGY AND MEDICINE 2021年第12期246卷 1355-1367页

作者： DiSpirito, Anthony, III Vu, Tri Pramanik, Manojit Yao, Junjie Duke Univ Dept Biomed Engn Durham NC 27708 USA Nanyang Technol Univ Sch Chem & Biomed Engn Singapore 637459 Singapore

The rapidly evolving field of photoacoustic tomography utilizes endogenous chromophores to extract both functional and structural information from deep within tissues. It is this power to perform precise quantitative measurements in vivo-with endogenous or exogenous contrast-that makes photoacoustic tomography highly promising for clinical translation in functional brain imaging, early cancer detection, real-time surgical guidance, and the visualization of dynamic drug responses. Considering photoacoustic tomography has benefited from numerous engineering innovations, it is of no surprise that many of photoacoustic tomography's current cutting-edge developments incorporate advances from the equally novel field of artificial intelligence. More specifically, alongside the growth and prevalence of graphical processing unit capabilities within recent years has emerged an offshoot of artificial intelligence known as deep learning. Rooted in the solid foundation of signal processing, deep learning typically utilizes a method of optimization known as gradient descent to minimize a loss function and update model parameters. There are already a number of innovative efforts in photoacoustic tomography utilizing deep learning techniques for a variety of purposes, including resolution enhancement, reconstruction artifact removal, undersampling correction, and improved quantification. Most of these efforts have proven to be highly promising in addressing long-standing technical obstacles where traditional solutions either completely fail or make only incremental progress. This concise review focuses on the history of applied artificial intelligence in photoacoustic tomography, presents recent advances at this multifaceted intersection of fields, and outlines the most exciting advances that will likely propagate into promising future innovations.

关键词： Photoacoustic tomography deep learning convolutional neural networks artificial intelligence photoacoustic computed tomography photoacoustic microscopy

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：