检索结果-内蒙古大学图书馆

A video object removal method based on pointwise spatial attention mechanism

学校读者我要写书评

暂无评论

A video object removal method based on pointwise spatial att...

2024 International Conference on Image, Signal Processing, and Pattern Recognition, ISPP 2024

作者： Li, Zhanli Wang, Daren Mu, Qi School of Computer Science and Technology Xi'an University of Science and Technology Xi'an China

ISBN: (纸本)9781510680425

In the advancing fields of computer vision and deep learning, video dynamic object removal technology is an important research area. The goal of this technology is to remove specific dynamic objects from a continuous video sequence while filling in the removed background. This paper will explore a new video dynamic object removal method based on feature pyramid and point by point spatial attention mask extraction network. This method can detect dynamic objects and fill the holes caused by dynamic object removal in the video. We have designed and implemented an end-to-end dynamic target removal system based on the above methods. The system first applies RCNN for dynamic target detection and generates corresponding masks. Then, a generative adversarial network is used to complete video restoration of the removed target area. In addition, the system focuses on optimizing performance to ensure processing speed and result quality. The experimental results of this article indicate that this method has high accuracy and effectiveness in removing and repairing dynamic targets in videos. This indicates that this new visual technology method has broad rule-based capabilities in handling moving objects and repairing their backgrounds. In the future, this technology can be widely applied in various video processing and repair projects, such as video editing, security monitoring, and other fields. © 2024 SPIE.

关键词： Generative adversarial networks

A Multi-Resolution Feature Fusion Method for Pedestrian Re-identification

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第9期51卷 1393-1403页

作者： Qin, Haitian Xu, Yang Chen, Xupeng School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

Pedestrian re-identification technology enables accurate identification of individuals and is widely used in modern intelligent video surveillance systems to aid law enforcement, including criminal apprehension and locating missing persons. However, variations in lighting, background, resolution, and other imaging conditions captured by different cameras create significant challenges in pedestrian feature extraction, often leading to poor recognition accuracy. To overcome these challenges, this paper presents a Multi-resolution Feature Fusion (MRFF) method for pedestrian re-identification, based on the Pedestrian Re-identification Relational Network (RNFPR). This approach incorporates the Coordinate Attention (CA) module into the DenseNet161 network to enhance feature extraction capabilities. Improving the discriminative and recognition accuracy of features requires learning and fusing pedestrian features from multiple low-resolution images. This process enhances the expressive power of feature maps, ultimately improving pedestrian recognition performance. Additionally, this method introduces a multi-resolution feature fusion module that segments and integrates multi-resolution features from image data. This enables the model to effectively combine feature information from various resolution levels, resulting in a more comprehensive feature representation. Experimental results show that the MRFF method achieves a 1.3% increase in mean Average Precision (mAP) and a 1.0% improvement in Rank-1 accuracy on the Market1501 dataset. For the DukeMTMC-reID dataset, it provides a 0.2% increase in mAP and a 0.7% enhancement in Rank-1 accuracy. Consequently, the MRFF approach results in an overall mAP increase of 0.7% on the DukeMTMC-ReID dataset, significantly improving pedestrian gender re-identification accuracy. © (2024), (International Association of Engineers). All Rights Reserved.

关键词： Security systems

Person Re-Identification Algorithm Based on Improved ResNet

学校读者我要写书评

暂无评论

IAENG International Journal of Applied Mathematics 2024年第5期54卷 894-901页

作者： Shen, Wenrui Wang, Zhifeng School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

Person Re-Identification falls within the scope of computer vision, acting a technique to ascertain the presence of a specified pedestrian within a video or image library. The related research is of great significance in real-world environments such as criminal investigation and statistical analysis of commercial foot traffic and has received extensive attention from the academic community. However, traditional methods such as manual extraction cannot adapt to large-scale data volumes, and deep learning-based methods at this stage suffer from interference in complex environments such as similar costumes, perspective changes, and occlusion. Therefore, in this paper, we investigate the above problems. Firstly, we expand the dataset by introducing random erasure-based preprocessing of pedestrian images to enhancing the robustness and generalization capability of neural networks. Secondly, a composite attention mechanism is introduced after the network residual layer to enhance the spatial information capability and feature expression. Finally, the union loss composed of Circle Loss, Ternary Loss, and Cross Entropy Loss was chosen for network training in the loss optimization phase. Findings from the experiments reveal that the improved method proposed in this experiment achieves 96.0%Rank-1 and 88.3%mAP in Market1501, which reflects the validity of the approach proposed in this manuscript, and provides valuable reference suggestions for Person Re-Identification related research. © (2024), (International Association of Engineers). All Rights Reserved.

关键词： Image enhancement

Improved Infrared Road Object Detection Algorithm Based on Attention Mechanism in YOLOv8

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第6期51卷 673-680页

作者： Luo, Zilong Tian, Ying School of Computer Science and Software Engineering University of Science and Technology Liaoning Liaoning114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

In Currently, research in the field of infrared road object detection is primarily focused on enhancing model performance and robustness to address the challenges posed by complex real-world driving scenarios. In response to these challenges, this paper proposes an infrared road object detection algorithm based on an attention mechanism. By incorporating the CPCA module, which utilizes attention mechanisms, into the YOLOv8s model, the algorithm enhances the model’s focus on unobstructed areas and highly illuminated sections, extracting crucial feature information to improve both accuracy and robustness. Additionally, the original model’s downsampling layer is replaced with the Context Grided Network Block Downsampling (CGBD) module, which not only preserves feature edge information but also effectively handles local and contextual features, thereby enhancing the overall feature capturing capabilities of the model. To address the issue of equal aspect ratios in the model’s original loss function, the proposed algorithm adopts the superior Weighted Intersection over Union (WIoU). This not only addresses the shortcomings of the original loss function (CIoU) but also demonstrates increased sensitivity in classification tasks. Experimental results show that the improved algorithm, compared to YOLOv8s, achieves a 1.4% increase in mean average precision (mAP), along with notable improvements in precision and recall. Furthermore, when compared to mainstream model algorithms, the enhanced model significantly outperforms in infrared road object detection tasks, providing validation of its effectiveness. © (2024), (International Association of Engineers). All rights reserved.

关键词： Object detection

Downhole Image Enhancement Algorithm Based on Improved CycleGAN 5

学校读者我要写书评

暂无评论

Downhole Image Enhancement Algorithm Based on Improved Cycle...

5th International Conference on computer Vision, Image and Deep Learning, CVIDL 2024

作者： Li, Na Gao, Sheng Xue, Jiameng Zhang, Yilong Xi'an University of Science and Technology School of Computer Science and Technology Xi'an China

ISBN: (纸本)9798350373820

Due to the complex environment of underground coal mines, the quality of captured images is often poor. The dim light in the underground leads to low image contrast and loss of detailed features;meanwhile, the dust and water vapor in the underground makes the image less clear and unevenly bright and dark. In addition, due to the small number of relevant datasets in the underground, it is unable to meet the requirements of deep learning algorithms, which causes great trouble to the underground image enhancement. To address the above problems, this paper proposes an image enhancement algorithm based on CycleGAN network. To address the problem of difficulty in acquiring paired image data in downhole, CycleGAN network is used for label-free training to expand the dataset;to address the problem of loss of image details in downhole, the improved UNET network DCSAU-Net is used instead of the original ResNet network to better preserve the image detail features and structure;by introducing self-attention in the discriminator of CycleGAN mechanism in the discriminator of CycleGAN to further improve the attention and reconstruction ability of important image information;finally, the EM distance with penalty term is used to replace the JS distance to alleviate the difficult problem that the CycleGAN network is not easy to converge. The experimental results show that compared with the original CycleGAN, the algorithm in this paper improves 6.281%, 15.714%, 7.627%, and 15.670% in the four indexes of PSNR, SSIM, IFC, and VIF, respectively. In terms of the visual effect of the images, the improved CycleGAN network generates images with higher clarity and better quality, which proves that the algorithm in this paper can effectively realize the low illumination image enhancement of coal mine underground. © 2024 IEEE.

关键词： Image enhancement

Analysis of radiation diffusion of COVID-19 driven by social attributes

学校读者我要写书评

暂无评论

Chinese Physics B 2024年第1期33卷 773-779页

作者：年福忠杨晓晨师亚勇 Lanzhou University of Technology School of Computer&CommunicationLanzhou 730050China University of Science and Technology Beijing School of Computer&Communication EngineeringBeijing 100083China

This paper first estimated the infectious capacity of COVID-19 based on the time series evolution data of confirmed cases in multiple countries. Then, a method to infer the cross-regional spread speed of COVID-19 was introduced in this paper, which took the gross domestic product(GDP) of each region as one of the factors that affect the spread speed of COVID-19 and studied the relationship between the GDP and the infection density of each region(China's Mainland, the United States, and EU countries). In addition, the geographic distance between regions was also considered in this method and the effect of geographic distance on the spread speed of COVID-19 was studied. Studies have shown that the probability of mutual infection of these two regions decreases with increasing geographic distance. Therefore, this paper proposed an epidemic disease spread index based on GDP and geographic distance to quantify the spread speed of COVID-19 in a region. The analysis results showed a strong correlation between the epidemic disease spread index in a region and the number of confirmed cases. This finding provides reasonable suggestions for the control of epidemics. Strengthening the control measures in regions with higher epidemic disease spread index can effectively control the spread of epidemics.

关键词： COVID-19 basic reproduction number gross domestic product(GDP) geographic distance cross-regional spread

Underwater Target Detection Based on Improved YOLOv7

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第4期51卷 422-429页

作者： Fu, Junshang Tian, Ying School of Computer Science and Software Engineering University of Science and Technology Liaoning Liaoning 114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan 114051 China

Underwater target detection is an important part of marine exploration. However, in complex underwater environments due to factors like light absorption and scattering, as well as variations in water quality and clarity. These challenges result in inaccurate target feature extraction, sluggish detection speeds, and insufficient robustness in the detection methods. In order to address these issues, an enhanced YOLOv7 network (YOLOv7-SPNW-D) is proposed for underwater target detection in this study. The SPD-MP module structure replaces the MP module in the neck network to capture small targets and enhance detection accuracy. A novel NWD loss function is employed to facilitate smoother extraction of small target features. This enhances feature extraction and improves network inference speed. Additionally, incorporating a small target detection module enables the providing of more comprehensive small target information within a deep feature map. This, in turn, improves the capture of small target features in complex backgrounds, and avoids feature loss and enhancing model exactness. Through ablation experiments on the URPC dataset, it is shown that the improved YOLOv7-SPNW-D algorithm performs better than the original YOLOv7 algorithm, with the mAP50 value increased to 87.0%, proving the effectiveness of this method. In conclusion, the improved YOLOv7-SPNW-D model is more suitable for underwater marine organism target detection. © (2024) International Association of Engineers.

关键词： Water quality

Small Object Detection in Aerial Drone Imagery based on YOLOv8

学校读者我要写书评

暂无评论

IAENG International Journal of computer science 2024年第9期51卷 1346-1354页

作者： Pan, Junyu Zhang, Yujun School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China School of Computer Science and Software Engineering University of Science and Technology Liaoning Anshan114051 China

In recent years, the utilization of unmanned aerial vehicles (UAVs) for aerial target detection has gained significant attention due to their high-altitude perspective and maneuverability, which offer novel opportunities and tremendous potential in this field. However, detecting targets in UAV aerial images remains highly challenging due to the presence of numerous small targets with limited feature information, as well as issues like target occlusion and complex backgrounds that severely impact detection accuracy. To address these challenges, we propose a detection model called BDC-YOLOv8 that aims to enhance accuracy for small targets while minimizing computational complexity. Specifically, we augment the YOLOv8 architecture by incorporating a dedicated detection head tailored for small targets to improve performance when encountering such objects. Additionally, we restructure the neck network of the model to better extract and fuse feature information from targets with significant scale variations. Furthermore, we introduce the concept of DynamicHead to enhance the detection head by incorporating various attention mechanisms suitable for our task ahead of the original detection head, thereby enhancing the model’s capability to detect objects of different scales and complex backgrounds. Moreover, we introduce Convolutional Block Attention Module (CBAM) to identify regions of interest in densely populated areas. Extensive experiments conducted on the VisDrone2019 dataset yield promising results where our model achieves a mean Average Precision (mAP) score of 38% and an AP50 score of 59.6%. Compared to the original YOLOV8 model, improvements are observed with increases in mAP by 2.5% and AP50 by 3.7%, respectively. Notably, our model demonstrates a significant enhancement in detecting small targets with an increase in APs evaluation metric by 4.1%. © (2024), (International Association of Engineers). All Rights Reserved.

关键词： Aerial photography

Multimodal Large Models are Effective Action Anticipators

学校读者我要写书评

暂无评论

IEEE Transactions on Multimedia 2025年 27卷 2949-2960页

作者： Wang, Binglu Tian, Yao Wang, Shunzhou Yang, Le Xi'an University of Architecture and Technology College of Information and Control Engineering Xi'an710055 China Northwestern Polytechnical University School of Astronautics Xi'an710072 China Peking University School of Electronic and Computer Engineering Shenzhen Graduate School Shenzhen518055 China Hefei Comprehensive National Science Center Institute of Artificial Intelligence Hefei230022 China University of Science and Technology of China School of Information Science and Technology Hefei230022 China

The task of long-term action anticipation demands solutions that can effectively model temporal dynamics over extended periods while deeply understanding the inherent semantics of actions. Traditional approaches, which primarily rely on recurrent units or Transformer layers to capture long-term dependencies, often fall short in addressing these challenges. Large Language Models (LLMs), with their robust sequential modeling capabilities and extensive commonsense knowledge, present new opportunities for long-term action anticipation. In this work, we introduce the ActionLLM framework, a novel approach that treats video sequences as successive tokens, leveraging LLMs to anticipate future actions. Our baseline model simplifies the LLM architecture by setting future tokens, incorporating an action tuning module, and reducing the textual decoder layer to a linear layer, enabling straightforward action prediction without the need for complex instructions or redundant descriptions. To further harness the commonsense reasoning of LLMs, we predict action categories for observed frames and use sequential textual clues to guide semantic understanding. In addition, we introduce a Cross-Modality Interaction Block, designed to explore the specificity within each modality and capture interactions between vision and textual modalities, thereby enhancing multimodal tuning. Extensive experiments on benchmark datasets demonstrate the superiority of the proposed ActionLLM framework, encouraging a promising direction to explore LLMs in the context of action anticipation. © 1999-2012 IEEE.

关键词： Semantics