检索结果-内蒙古大学图书馆

GazeREC-Net: Advancing Gaze Restoration in Low-Light Conditions

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第12期51卷 2034-2042页

作者： Ku, Jiayin Wang, Li School of Computer Science and software Engineering University of Science and Technology Liaoning Anshan114051 China College of Computer Science and Tech nology Liaoning Anshan114051 China

Gaze estimation technology is essential for applications such as human-computer interaction, augmented reality, and virtual reality. However, its accuracy is significantly compromised in low-light conditions due to degraded image quality. To address this, we developed GazeREC-Net, an innovative gaze restoration method. We simulated low-light conditions on the MPIIFaceGaze and ColumbiaGaze datasets, creating a specialized degraded dataset for training and testing our model. GazeREC-Net combines Fourier transform techniques with advanced image restoration algorithms, significantly enhancing image quality and optimizing gaze information recovery and extraction in low-light environments. Our evaluations using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) metrics demonstrated a PSNR improvement of 7.64% and an SSIM improvement of 5.75% compared to HINet. Additionally, GazeREC-Net outperformed CFTNet and other existing gaze estimation models, including Gaze-TR, Dilated-Net, CA-Net, L2CS-Net, and MTGLS, in reducing gaze estimation error. These findings validate the effectiveness of GazeREC-Net in low-light conditions and offer new research directions for applying gaze estimation technologies in complex lighting environments. © (2024), (International Association of Engineers). All rights reserved.

关键词： Augmented reality

An Intelligent Privacy Protection Scheme for Efficient Edge Computation Offloading in IoV

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2024年第4期33卷 910-919页

作者： Liang YAO Xiaolong XU Wanchun DOU Muhammad Bilal School of Software Nanjing University of Information Science and Technology State Key Laboratory for Novel Software Technology Nanjing University Department of Computer and Electronics Systems Engineering Hankuk University of Foreign Studies

As a pivotal enabler of intelligent transportation system(ITS), Internet of vehicles(Io V) has aroused extensive attention from academia and industry. The exponential growth of computation-intensive, latency-sensitive,and privacy-aware vehicular applications in Io V result in the transformation from cloud computing to edge computing,which enables tasks to be offloaded to edge nodes(ENs) closer to vehicles for efficient execution. In ITS environment,however, due to dynamic and stochastic computation offloading requests, it is challenging to efficiently orchestrate offloading decisions for application requirements. How to accomplish complex computation offloading of vehicles while ensuring data privacy remains challenging. In this paper, we propose an intelligent computation offloading with privacy protection scheme, named COPP. In particular, an Advanced Encryption Standard-based encryption method is utilized to implement privacy protection. Furthermore, an online offloading scheme is proposed to find optimal offloading policies. Finally, experimental results demonstrate that COPP significantly outperforms benchmark schemes in the performance of both delay and energy consumption.

关键词： Industries Privacy Energy consumption Transportation Computational efficiency Encryption Protection

DA-FR-UNet: A Dual Attention-Enhanced Approach for Retinal Vessel Image Segmentation

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2025年第5期52卷 1325-1335页

作者： Li, Yuanyuan Wu, Jiansheng Liu, Yang School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China University of Science and Technology Liaoning Anshan114051 China

Retinal vessel image segmentation is crucial for the early diagnosis and treatment of ophthalmic diseases. However, accurate segmentation remains challenging due to the complexity and diversity of retinal images, as well as the intricate nature of vascular structures. Existing methods often underperform when dealing with fine vessels and complex backgrounds, limiting their effectiveness in clinical applications. This study proposes a novel retinal vessel image segmentation method, Dual Attention FR-UNet (DA-FR-UNet), which incorporates two innovative attention mechanisms: Enhanced Channel Prior Attention (ECPA) and Hard Triplet Attention (HTA). The ECPA module enhances the model's ability to recognize fine blood vessels by integrating global average pooling and global maximum pooling to effectively capture multi-scale features. In contrast, the HTA module improves the model's understanding of complex vascular structures by simultaneously considering attention information across three dimensions: channel, spatial height, and width. Additionally, the study introduces the testtime augmentation (TTA) technique during the experimental phase to enhance the model's generalizability and robustness. To validate the performance of the DA-FR-UNet model in retinal vessel image segmentation tasks, a series of comparative experiments were conducted on the FIVES and RVD datasets. The experimental results demonstrate that the DA-FR-UNet model, combining the dual-attention mechanism and the TTA strategy, outperforms existing state-of-the-art methods across several key evaluation metrics. On the FIVES dataset, DAFR- UNet achieves a sensitivity of 88.42%, an IoU value of 82.27%, and a Dice coefficient of 89.72%. On the RVD dataset, these metrics are 65.46%, 44.53%, and 61.43%, respectively. These findings highlight the model's significant advantages in processing fine vessels and complex backgrounds, improving its sensitivity and robustness in blood vessel image segmentation. © (2025)

关键词： Image segmentation

Active self-training for weakly supervised 3D scene semantic segmentation

学校读者我要写书评

暂无评论

Computational Visual Media 2024年第3期10卷 425-438页

作者： Gengxin Liu Oliver van Kaick Hui Huang Ruizhen Hu College of Computer Science&Software Engineering Shenzhen UniversityShenzhen 518060China School of Computer Science Carleton UniversityOttawa K1S 5B6Canada

Since the preparation of labeled datafor training semantic segmentation networks of pointclouds is a time-consuming process, weakly supervisedapproaches have been introduced to learn fromonly a small fraction of data. These methods aretypically based on learning with contrastive losses whileautomatically deriving per-point pseudo-labels from asparse set of user-annotated labels. In this paper, ourkey observation is that the selection of which samplesto annotate is as important as how these samplesare used for training. Thus, we introduce a methodfor weakly supervised segmentation of 3D scenes thatcombines self-training with active learning. Activelearning selects points for annotation that are likelyto result in improvements to the trained model, whileself-training makes efficient use of the user-providedlabels for learning the model. We demonstrate thatour approach leads to an effective method that providesimprovements in scene segmentation over previouswork and baselines, while requiring only a few userannotations.

关键词： semantic segmentation weakly supervised self-training active learning

Research on Enhanced Multi-head Self-Attention Social Recommendation Algorithm Based on Graph Neural Network

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第7期51卷 754-764页

作者： Teng, Yue Yang, Kai Yue Teng is a postgraduate student at the School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

In today’s society, communication among people has become more frequent and extensive due to the rapid development of science, technology, and the Internet. This vast communication occurs in real life and virtual online worlds. However, with the overwhelming amount of information available, people often struggle with finding and accessing the information they need, leading to frustration. To address this challenge, we conducted a study to extract valuable information and insights from users’ social relationships. While existing models can help us understand and analyze users’ social relationships to some extent, there are still significant challenges in extracting valuable insights. Therefore, we conducted in-depth research to improve the GraphRec model and created a new social recommendation model called the CWYGNN model. The CWYGNN model is unique because it can identify deeper connections in users’ social interactions through in-depth learning of the social sequences between users. This is important for understanding users’ behavioral patterns, predicting their needs, and providing personalized services. To validate the effectiveness of the CWYGNN model, we conducted experiments and comparisons on several public datasets. The experimental results show that the CWYGNN model outperforms similar approaches in processing user social relationships and extracting valuable insights. This result provides a new way of thinking and methodology to help us better understand and serve users and ultimately enhance their overall experience. © (2024) International Association of Engineers. All rights reserved.

关键词： Recommender systems

Diabetic Retinopathy Image Segmentation Method Based on Fusion DenseNet and U-Net Network

学校读者我要写书评

暂无评论

engineering Letters 2025年第2期33卷 418-428页

作者： Zhang, Yaxuan Xu, Yang Zhai, Jinlong University of Science and Technology LiaoNing Anshan114051 China School of Computer Science and Software Engineering University of Science and Technology LiaoNing Anshan114051 China

Diabetic Retinopathy (DR) is a common and significant complication in patients with diabetes, and severely affecting their quality of life. Image segmentation plays a crucial role in the early diagnosis and treatment of DR. However, traditional methods are limited in terms of segmentation accuracy and generalization capability. This paper proposes a novel image segmentation method for diabetic retinopathy based on the integration of DenseNet and U-Net networks. Firstly, DenseNet is utilized to replace the encoder part of U-Net. The dense connection mechanism enhances the efficiency of feature propagation and improves the feature extraction capability of the encoder. Secondly, we introduce Omni-Dimensional Dynamic Convolution (ODConv) to replace traditional convolutions. ODConv is used to handle diverse input features effectively. The model's adaptability and segmentation accuracy for various samples is improved. Finally, we integrate the Convolutional Block Attention Module (CBAM) into the decoder. By leveraging channel and spatial attention, the ability to capture key features is enhanced, and the segmentation accuracy and boundary recognition capabilities are improved. To validate the effectiveness and feasibility of the model. This paper conducts experiments on the DDR dataset and the IDRID dataset. The experimental results show that the proposed method improved the mDice score by 0.53% compared to traditional methods. It also increased the mIoU score by 1.12% over traditional methods. This indicates that the proposed method has better performance in segmenting diabetic retinopathy images. It also demonstrates better generalization ability. © 2025, International Association of Engineers. All rights reserved.

关键词： Image segmentation

Infrared Action Recognition Model Based on Improved ST-GCN

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2025年第6期52卷 1664-1671页

作者： Zhu, Xiaoliang Zhou, Ziwei School of Electronic Information University of Science and Technology Liaoning Anshan114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

Infrared imaging technology is capable of capturing the thermal radiation emitted by the human body in conditions with insufficient visible light. Consequently, infrared behavior recognition leverages this capability to detect and analyze human movements in low-light or complex environments. However, infrared images are often affected by noise interference, which can obscure target features. To tackle these challenges, we propose an infrared human behavior recognition model. Within this model, human regions in infrared images are detected by YOLOv8 and passed on to AlphaPose to predict the locations of skeletal keypoints in the human body. Subsequently, the acquired skeletal sequences are employed to predict actions in ST-GCN. Simultaneously, we introduced the LKA attention mechanism and the PReLU activation function for structural optimization within the ST-GCN. These improvements enabled the ST-GCN to extract action features from skeletal keypoints more effectively, thereby enhancing the accuracy of infrared behavior *** extensive ablation studies, we have demonstrated that our proposed LPST-GCN model significantly enhances the performance of infrared action recognition and achieves excellent results on both the UNISV dataset (99.02%) and the NTU RGB+D dataset (95.86%). © (2025), IAENG International Journal of computer science

关键词： Thermography (imaging)

STAL3D: Unsupervised Domain Adaptation for 3D Object Detection via Collaborating Self-Training and Adversarial Learning

学校读者我要写书评

暂无评论

IEEE Transactions on Intelligent Vehicles

IEEE Transactions on Intelligent Vehicles 2024年第11期9卷 1-12页

作者： Zhang, Yanan Zhou, Chao Huang, Di State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beihang University Beijing China

Existing 3D object detection suffers from expensive annotation costs and poor transferability to unknown data due to the domain gap, Unsupervised Domain Adaptation (UDA) aims to generalize detection models trained in labeled source domains to perform robustly on unexplored target domains, providing a promising solution for cross-domain 3D object detection. Although Self-Training (ST) based cross-domain 3D detection methods with the assistance of pseudo-labeling techniques have achieved remarkable progress, they still face the issue of low-quality pseudo-labels when there are significant domain disparities due to the absence of a process for feature distribution alignment. While Adversarial Learning (AL) based methods can effectively align the feature distributions of the source and target domains, the inability to obtain labels in the target domain forces the adoption of asymmetric optimization losses, resulting in a challenging issue of source domain bias. To overcome these limitations, we propose a novel unsupervised domain adaptation framework for 3D object detection via collaborating ST and AL, dubbed as STAL3D, unleashing the complementary advantages of pseudo labels and feature distribution alignment. Additionally, a Background Suppression Adversarial Learning (BS-AL) module and a Scale Filtering Module (SFM) are designed tailored for 3D cross-domain scenes, effectively alleviating the issues of the large proportion of background interference and source domain size bias. Our STAL3D achieves state-of-the-art performance on multiple cross-domain tasks and even surpasses the Oracle results on Waymo $\rightarrow$ KITTI and Waymo $\rightarrow$ KITTI-rain. IEEE

关键词： Feature extraction

Optimizing Risk-Aware Task Migration Algorithm Among Multiplex UAV Groups Through Hybrid Attention Multi-Agent Reinforcement Learning

学校读者我要写书评

暂无评论

Tsinghua science and Technology 2025年第1期30卷 318-330页

作者： Yuanshuang Jiang Kai Di Ruiyi Qian Xingyu Wu Fulin Chen Pan Li Xiping Fu Yichuan Jiang School of Computer Science and Engineering Southeast UniversityNanjing 211189 School of Software Engineering Southeast UniversityNanjing 211189 School of Cyber Science and Engineering Southeast UniversityNanjing 211189 PredictHQ Auckland 1010New Zealand

Recently,with the increasing complexity of multiplex Unmanned Aerial Vehicles(multi-UAVs)collaboration in dynamic task environments,multi-UAVs systems have shown new characteristics of inter-coupling among multiplex groups and intra-correlation within ***,previous studies often overlooked the structural impact of dynamic risks on agents among multiplex UAV groups,which is a critical issue for modern multi-UAVs communication to *** address this problem,we integrate the influence of dynamic risks on agents among multiplex UAV group structures into a multi-UAVs task migration problem and formulate it as a partially observable Markov *** then propose a Hybrid Attention Multi-agent Reinforcement Learning(HAMRL)algorithm,which uses attention structures to learn the dynamic characteristics of the task environment,and it integrates hybrid attention mechanisms to establish efficient intra-and inter-group communication aggregation for information extraction and group *** results show that in this comprehensive and challenging model,our algorithm significantly outperforms state-of-the-art algorithms in terms of convergence speed and algorithm performance due to the rational design of communication mechanisms.

关键词： Unmanned Aerial Vehicle(UAV) multiplex UAV group structures task migration multi-agent reinforcement learning