检索结果-内蒙古大学图书馆

Monocular 3D Object Detection Utilizing Auxiliary Learning With Deformable Convolution

IEEE Transactions on Intelligent Transportation Systems 2024年第3期25卷 2424-2436页

作者： Chen, Jiun-Han Shieh, Jeng-Lun Haq, Muhamad Amirul Ruan, Shanq-Jang National Taiwan University of Science and Technology Department of Electronic and Computer Engineering Taipei10607 Taiwan

In autonomous driving systems, the monocular 3D object detection algorithm is a crucial component. The safety of autonomous vehicles heavily depends on a well-designed detection system. Therefore, developing a robust and efficient 3D object detection algorithm is a major goal for institutes and researchers. Having a 3D sense is essential in autonomous vehicles and robotics, as it allows the system to understand its surroundings and react accordingly. Compared with stereo-based and Lidar-based methods, monocular 3D Object detection is a challenging task as it only utilizes 2D information to generate complex 3D features, making it low-cost, less computationally intensive, and with great potential. However, the performance of monocular methods is impaired due to the lack of depth information. In this paper, we propose a simple, end-to-end, and effective network for monocular 3D object detection without the use of external training data. Our work is inspired by auxiliary learning, in which we use a robust feature extractor as our backbone and multiple regression heads to learn auxiliary knowledge. These auxiliary regression heads will be discarded after training for improved inference efficiency, allowing us to take advantage of auxiliary learning and enabling the model to learn critical information more conceptually. The proposed method achieves 17.28% and 20.10% for the moderate level of the Car category on the KITTI benchmark test set and validation set, respectively, which outperforms the previous monocular 3D object detection approaches. © 2000-2011 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

3D Structure Estimation of Room Environment Using Semantic Segmentation and Model Fitting

3D Structure Estimation of Room Environment Using Semantic S...

引用

2024 IEEE International Workshop on Metrology for Living Environment, MetroLivEnv 2024

作者： Morioka, Junya Miyamoto, Ryusuke Graduate School of Science and Technology Meiji University Department of Computer Science Kawasaki Japan School of Science and Technology Meiji University Department of Computer Science Kawasaki Japan

ISBN: (纸本)9798350385014

Recent advances in VR devices and technology have increased the demand for technologies that reflect real space in virtual space. In particular, technologies that reflect the indoor environment in a virtual space are used in various fields such as architecture, interior design, and robotics. To reflect a real space in a virtual space, estimating the 3D structure of the real space is significant. Yet, this requires expensive equipment and poses challenges in computational complexity and ease of measurement. In this study, we propose a method of estimating 3D structure of indoor environments using a single image. Existing methods generally use depth estimation, but in this study, we introduce a new approach using semantic segmentation and rectangular fitting. Experimental results with 3D CG show that the proposed method outperforms existing methods in accuracy and competitive results. © 2024 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A Mining Conveyor Belt Tear Detection Method based on Improved STDC 9

A Mining Conveyor Belt Tear Detection Method based on Improv...

引用

9th International Symposium on computer and Information Processing technology, ISCIPT 2024

作者： Wang, Zhen Li, Xiaoming Taiyuan University of Science and Technology Department of Computer Science and Technology Taiyuan China

ISBN: (纸本)9798350388404

Real-time and accurate detection of conveyor belt tears is a key issue in industrial applications, and this problem is still not well solved. In this paper, we propose an improved STDC network to enhance the performance of tear detection. Firstly, the STDC module in the backbone network is reorganized to enhance the feature extraction capability of the network. Secondly, the object filtering module is designed to capture the inter-channel dependencies while discarding useless information to reduce feature redundancy. Lastly, in the feature fusion stage, the fusion of high-level semantic information and low-level spatial information is strengthened by focusing on the channel and spatial features at the same time, which optimizes the output of the network. Experimental results show that detection performance of our method in complex environments is significantly improved with an FPS of 39.5 and an mIoU of 91.39%. Compared with other algorithms, our method achieves a better balance between segmentation speed and accuracy. © 2024 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Parallel Computation of Dominance Scores for Multidimensional Datasets on GPUs

引用

IEEE Transactions on Parallel and Distributed Systems 2024年第6期35卷 764-776页

作者： Chen, Wei-Mei Tsai, Hsin-Hung Ling, Joon Fong National Taiwan University of Science and Technology Department of Electronic and Computer Engineering Taipei106 Taiwan

The dominance scoring problem in a multidimensional dataset is to return the number of points dominated by a given point, which is a common metric for evaluating the quality of a data point. Dominance scoring is an elementary operator for variations of the skyline operator, including top k dominating and k-skyband queries. This study proposes query processing for dominance scores that operates primarily on the graphics processing unit (GPU) to fully utilize its massive processing resources and restricted memory space while reducing the transfer overhead between the central processing unit (CPU) and GPU. We introduce a heap-based multidimensional data structure with complete and well-balanced characteristics. Using our preprocessed data, we can construct a complete R-tree with the non-overlapping property, ensuring that the bounding boxes of internal nodes of the same level do not overlap, thereby reducing redundant operations. In addition, we propose two algorithms based on depth-first and breadth-first traversals to accumulate the dominance score on GPUs in parallel. Both take full advantage of the GPU's computing resources and memory space supported by the non-overlapping tree structures. Experiments on synthetic and real-world datasets demonstrate that the proposed algorithms implemented on GPUs dramatically improve the efficiency of dominance scoring. © 1990-2012 IEEE.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

VTM-GAN: video-text matcher based generative adversarial network for generating videos from textual description

引用

International Journal of Information technology (Singapore) 2024年第1期16卷 221-236页

作者： Mehmood, Rayeesa Bashir, Rumaan Giri, Kaiser J. Department of Computer Science Islamic University of Science and Technology Jammu and Kashmir 192122 India

Text-to-video synthesis has garnered significant attention as a challenging task in the domain of vision computing. With the advent of unsupervised learning techniques, text-to-video synthesis has become more feasible. In this context, Generative Adversarial Network (GAN)-based training networks have emerged as the leading unsupervised deep learning methods, exhibiting promising results. However, achieving visual quality, temporal coherence, and semantic consistency between the generated video and textual descriptions remains a considerable challenge. In this paper, we propose a novel approach called Video-Text Matcher (VTM) based GAN for text-to-video synthesis. The proposed VTM is based on Contrastive Language-Image Pre-training (CLIP) but with modifications. It incorporates both global sentence-level and fine-grained word-level information to calculate the similarity between the generated video and the provided textual descriptions. Unlike CLIP, which focuses on matching losses at the global sentence-image level only, our VTM includes a word-region level loss to enhance the fine granularity consistency between the text and video. We evaluate our proposed approach using the Single Digit Bouncing MNIST GIFs (SBMG) dataset and conduct both qualitative and quantitative analyses. The results demonstrate that our proposed method generates appealing videos that align well with the given textual descriptions, showcasing the effectiveness of our approach for text-to-video synthesis. © 2023, The Author(s), under exclusive licence to Bharati Vidyapeeth's Institute of computer Applications and Management.

关键词： Contrastive language–image pre-training Generative adversarial network Text-to-video generation Video-text matcher

来源：评论

学校读者我要写书评

暂无评论

Optimizing photo-to-anime translation with prestyled paired datasets

引用

Multimedia Tools and Applications 2024年第41期83卷 89393-89414页

作者： Chang, Chuan-Wang Dharmawan, Pratamagusta Department of Computer Science and Information Engineering National Chin-Yi University of Technology Taichung Taiwan

Animation is a widespread artistic expression that holds a special place in people's hearts. Traditionally, animation creation has relied heavily on manual techniques, demanding skilled drawing abilities and a significant amount of time. For instance, many Japanese anime films draw inspiration from real-world settings, requiring access to relevant references and artists capable of translating them into anime visuals. Consequently, the development of technology that automatically converts images into anime holds great significance. Numerous methods for style transfer have been developed using unsupervised learning and have achieved impressive results. However, unsupervised learning methods suffer when the image contains multiple styles within itself because they learn the style of the image globally. To solve this problem, we propose splitting these styles within the image into multiple classes: sky, buildings, greenery, water, and other objects. Then, we style these separated classes using existing image-to-image translation models. Finally, we train a pix2pix model to learn image style transfer in a paired manner. The experimental results show that the images are effectively styled into the resulting anime-styled image domain with comparable results to existing unsupervised learning GAN-based methods. The proposed method can effectively transfer the style from real-world photos into the anime-styled image domain. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

AMCB: enhancing the authentication process with blockchain integrated with PUF and machine learning

引用

Multimedia Tools and Applications 2025年第8期84卷 4809-4835页

作者： Al-Ghuraybi, Hind A. AlZain, Mohammed A. Soh, Ben Department of Information Technology College of Computers and Information Technology Taif University Taif21944 Saudi Arabia Department of Computer Science and Computer Engineering La Trobe University Bundoora Australia

The rapid advancement of technology has given rise to medical cyber-physical systems (MCPS), a subset of cyber-physical systems (CPS) specifically tailored for patient care and healthcare providers. MCPS generate substantial volumes of big data, posing challenges in terms of processing, storage, and safeguarding against unauthorized access. This paper introduces the Authenticated Medical Cyber-Physical Blockchain (AMCB) model within MCPS, leveraging Blockchain technology, machine learning, and physically unclonable functions (PUF) to fortify the authentication process, particularly for accessing electronic health records (EHR) stored on cloud servers. The model also involves analyzing data from authenticated devices to ensure its integrity. Experiments on the AMCB model evaluate its efficiency across four layers. In the physical layer, PUF keys are extracted using a built Static Random-Access Memory (SRAM) circuit and stored in a simulated Blockchain, yielding a Hamming Distance standard deviation of 6.60%. The inspection layer analyzes data from authenticated devices using machine learning, with the Random Forest achieving the highest accuracy at 99.21%. In the application layer, centered on user authentication, the average transaction time is 36.15 MS. Overall, the AMCB model requires an average of 49.82 ms to execute operations across all layers. These findings underscore the effectiveness of the proposed model in securing EHR in the dynamic landscape of MCPS. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Authentication

来源：评论

学校读者我要写书评

暂无评论

Reinforcement learning (RL)-based semantic segmentation and attention based backpropagation convolutional neural network (ABB-CNN) for breast cancer identification and classification using mammogram images

引用

Neural Computing and Applications 2024年第24期36卷 14797-14823页

作者： Thakur, Neha Kumar, Pardeep Kumar, Amit Department of Computer Science & amp Engineering Jaypee University of Information Technology HP Solan India

Breast cancer poses a threat to women’s health and contributes to an increase in mortality rates. Mammography has proven to be an effective tool for the early detection of breast cancer. However, it faces many challenges in early breast cancer detection due to poor image quality, traditional segmentation, and feature extraction. Therefore, this work addresses these issues and proposes an attention-based backpropagation convolutional neural network (ABB-CNN) to detect breast cancer from mammogram images more accurately. The proposed work includes image enhancement, reinforcement learning-based semantic segmentation (RLSS), and multiview feature extraction and classification. The image enhancement is performed by removing noise and artefacts through a hybrid filter (HF), image scaling through a pixel-based bilinear interpolation (PBI), and contrast enhancement through an election-based optimization (EO) algorithm. In addition, the RLSS introduces intelligent segmentation by utilizing a deep Q network (DQN) to segment the region of interest (ROI) strategically. Moreover, the proposed ABB-CNN facilitates multiview feature extraction from the segmented region to classify the mammograms into normal, malignant, and benign classes. The proposed framework is evaluated on the collected and the digital database for screening mammography (DDSM) datasets. The proposed framework provides better outcomes in terms of accuracy, sensitivity, specificity, precision, f-measure, false-negative rate (FNR) and area under the curve (AUC). This work achieved (99.20%, 99.35%), (99.56%, 99.66%), (98.96%, 98.99%), (99.05%, 99.12%), (0.44%, 0.34%), (99.31%, 99.39%) and (99.27%, 99.32%) of accuracy, sensitivity, specificity, precision, FNR, f-measure and AUC on (collected, DDSM datasets), respectively. This research addresses the prevalent challenges in breast cancer identification and offers a robust and highly accurate solution by integrating advanced deep-learning techniques. The evaluated re

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

DNN-STACK: a stacking technique based on deep neural network for detecting copy-move forgery

引用

Neural Computing and Applications 2025年第6期37卷 4989-5004页

作者： Krishnalal, G. Jagathy Raj, V.P. Madhu, G. Arun, K.S. School of Engineering Cochin University of Science and Technology Kerala Cochin682022 India School of Management Studies Cochin University of Science and Technology Kerala Cochin682022 India Department of Computer Applications Cochin University of Science and Technology Kerala Cochin682022 India

In recent years, detecting image forgery has become an important topic because of the availability of efficient and sophisticated image-editing software. Copy-move forgery, which involves copying a section of an image and pasting it to a different location inside the same image, is one of the most popular tampering techniques that alter the identity of a given image. This paper presents DNN-STACK, a deep neural network-based stacking scheme to identify copy-move forgery by effectively combining the predictions yielded by different base-level models. To extract hierarchical representations from the given images and generate base-level predictions, the proposed approach uses five distinct but complementary deep learning-based base-level models. After that, a consensus prediction is generated using the proposed DNN-STACK model, which can infer the nonlinear relationship between the base-level predictions. Experimental evaluation on three publicly available datasets, such as MICC-F600, MICC-F2000, and FAU, reveals the fact that the proposed DNN-STACK model outperforms the existing forgery detection techniques, significantly improving detection accuracy across varying image resolutions and attack types, including rotation, scaling, noise, and compression levels. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

MCI-GAN: a novel GAN with identity blocks inspired by menstrual cycle behavior for missing pixel imputation

引用

Neural Computing and Applications 2025年第16期37卷 9669-9703页

作者： Marie, Hanaa Salem Elbaz, Mostafa Faculty of Artificial Intelligence Delta University for Science and Technology Gamasa35712 Egypt Department of Computer Science Faculty of Computers and Informatics Kafrelsheikh University Kafrelsheikh Egypt

This paper presents MCI-GAN, a novel menstrual cycle imputation (MCI) and generative adversarial network (GAN) framework designed to address the challenge of missing pixel imputation in medical images. Inspired by the intelligent behavior of the endometrial lining during the menstrual cycle, our method introduces four key innovations. First, we propose a novel metaheuristic algorithm that assigns weights to surround pixels based on menstrual cycle behavior, ensuring that the imputed pixels maintain structural integrity and coherence with their neighbors, thus preserving overall image quality. Second, to enhance the learning capability of the GAN, identity blocks are integrated into the network architecture, improving the network’s ability to capture complex spatial relationships and leading to more accurate and consistent imputation of missing pixels. Third, we introduce an adaptive loss function that dynamically adjusts the penalty for pixel discrepancies based on local image context, allowing the model to focus on areas where accurate imputation is most critical and thereby enhancing overall image fidelity. Fourth, the framework incorporates a multi-scale feature extraction mechanism, enabling the GAN to process and combine information at various levels of detail, ensuring that both fine-grained textures and larger structural patterns are accurately captured during the imputation process. The efficacy of MCI-GAN is demonstrated across three diverse medical imaging datasets: mammograms, magnetic resonance imaging (MRI) scans, and skin lesion images. Our results show that the proposed method significantly outperforms existing approaches in terms of accuracy and structural coherence, offering a robust solution for missing pixel imputation in medical imaging. © The Author(s) 2025.

关键词： Magnetic resonance imaging

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：