检索结果-内蒙古大学图书馆

Editorial: Efficient deep neural network for intelligent robot system: Focusing on visual signal processing

学校读者我要写书评

暂无评论

Frontiers in Neurorobotics 2023年 17卷 1191655页

作者： Bai, Xiao Ning, Xin Donta, Praveen Kumar Li, Weijun School of Computer Science and Engineering State Key Laboratory of Software Development Environment Jiangxi Research Institute Beihang University Beijing China Institute of Semiconductors Chinese Academy of Sciences Beijing China Vienna Austria

can be effectively applied to facial expression recognition. Lu et al. present a facial image inpainting method using a multistage GAN and the global attention mechanism. The proposed method can effectively restore incomplete facial images by enhancing feature mining and semantic expression, using skip connections, encoder-decoder structure, and a local refinement network. Comparative experiments demonstrate that the proposed method generates realistic inpainting results with high PSNR and SSIM, indicating the model's performance and efficiency. Lin et al. address the shortage of boxing coaches in Chinese campuses by proposing a novel solution that employs human pose estimation technology to train interns. Specifically, they develop a model transfer technique that utilizes channel patching to enhance the accuracy of pose key points by an average of 1-20% and 3D accuracies by 0.3-0.5% compared to 2D baselines. The proposed method is not only practical but also effective for boxing pose *** contributions focus on the implementation of lightweight models to address other signals. Zheng presents a writing feature abstraction process based on ON-LSTM and attention mechanism for sentiment analysis, addressing the problem of ignoring syntactic and tag semantics information in emotional text feature extraction. The study shows the high application potential of deep learning models for dynamic user sentiment analysis. Wang and Chen investigate teachers' acceptance of robotics education and its relationship to the effectiveness and sustainability of robotics education using the UTAUT model and deep learning algorithms. The study also found that deep learning models such as mDAE and AmDAE reduced training time compared to existing noise-reducing autoencoder models. Teng et al. address the lack of technical and algorithmic support in music therapy for cancer patients and design a neural network robotic system based on breast cancer patients to analyze the ef

关键词： deep neural network intelligent robot system image processing pattern recognition visual navigation

AdvSwap: Covert Adversarial Perturbation with High Frequency Info-swapping for Autonomous Driving Perception

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Huang, Yuanhao Zhang, Qinfan Xing, Jiandong Cheng, Mengyue Yu, Haiyang Ren, Yilong Xiong, Xiao School of Transportation Science and Engineering Beihang University Beijing100191 China State Key Lab of Intelligent Transportation System Beijing100191 China Zhongguancun Laboratory Beijing100094 China Department of Electrical and Computer Engineering University of Alberta Edmonton Canada

Perception module of Autonomous vehicles (AVs) are increasingly susceptible to be attacked, which exploit vulnerabilities in neural networks through adversarial inputs, thereby compromising the AI safety. Some researches focus on creating covert adversarial samples, but existing global noise techniques are detectable and difficult to deceive the human visual system. This paper introduces a novel adversarial attack method, AdvSwap, which creatively utilizes wavelet-based high-frequency information swapping to generate covert adversarial samples and fool the camera. AdvSwap employs invertible neural network for selective high-frequency information swapping, preserving both forward propagation and data integrity. The scheme effectively removes the original label data and incorporates the guidance image data, producing concealed and robust adversarial samples. Experimental evaluations and comparisons on the GTSRB and nuScenes datasets demonstrate that AdvSwap can make concealed attacks on common traffic targets. The generates adversarial samples are also difficult to perceive by humans and algorithms. Meanwhile, the method has strong attacking robustness and attacking transferability. © 2025, CC BY-NC-ND.

关键词：

Efficient Long-Short Temporal Attention Network for Unsupervised Video Object Segmentation

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Li, Ping Zhang, Yu Yuan, Li Xiao, Huaxin Lin, Binbin Xu, Xianghua School of Computer Science and Technology Hangzhou Dianzi University Hangzhou China State Key Laboratory for Novel Software Technology Nanjing University Nanjing China School of Electrical and Computer Engineering Peking University Beijing China College of Systems Engineering National University of Defense Technology Changsha China College of Computer Science Zhejiang University Hangzhou China

Unsupervised Video Object Segmentation (VOS) aims at identifying the contours of primary foreground objects in videos without any prior knowledge. However, previous methods do not fully use spatial-temporal context and fail to tackle this challenging task in real-time. This motivates us to develop an efficient Long-Short Temporal Attention network (termed LSTA) for unsupervised VOS task from a holistic view. Specifically, LSTA consists of two dominant modules, i.e., Long Temporal Memory and Short Temporal Attention. The former captures the long-term global pixel relations of the past frames and the current frame, which models constantly present objects by encoding appearance pattern. Meanwhile, the latter reveals the short-term local pixel relations of one nearby frame and the current frame, which models moving objects by encoding motion pattern. To speedup the inference, the efficient projection and the locality-based sliding window are adopted to achieve nearly linear time complexity for the two light modules, respectively. Extensive empirical studies on several benchmarks have demonstrated promising performances of the proposed method with high © 2023, The Authors. All rights reserved.

关键词： Pixels

CoR-GS: Sparse-View 3D Gaussian Splatting via Co-Regularization

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhang, Jiawei Li, Jiahe Yu, Xiaohan Huang, Lei Gu, Lin Zheng, Jin Bai, Xiao School of Computer Science and Engineering State Key Laboratory of Complex & Critical Software Environment Jiangxi Research Institute Beihang University China SKLCCSE Institute of Artificial Intelligence Beihang University China School of Computing Macquarie University Australia RIKEN AIP Japan The University of Tokyo Japan

3D Gaussian Splatting (3DGS) creates a radiance field consisting of 3D Gaussians to represent a scene. With sparse training views, 3DGS easily suffers from overfitting, negatively impacting rendering. This paper introduces a new co-regularization perspective for improving sparse-view 3DGS. When training two 3D Gaussian radiance fields, we observe that the two radiance fields exhibit point disagreement and rendering disagreement that can unsupervisedly predict reconstruction quality, stemming from the randomness of densification implementation. We further quantify the two disagreements and demonstrate the negative correlation between them and accurate reconstruction, which allows us to identify inaccurate reconstruction without accessing ground-truth information. Based on the study, we propose CoR-GS, which identifies and suppresses inaccurate reconstruction based on the two disagreements: (i) Co-pruning considers Gaussians that exhibit high point disagreement in inaccurate positions and prunes them. (ii) Pseudo-view co-regularization considers pixels that exhibit high rendering disagreement are inaccurate and suppress the disagreement. Results on LLFF, Mip-NeRF360, DTU, and Blender demonstrate that CoR-GS effectively regularizes the scene geometry, reconstructs the compact representations, and achieves state-of-the-art novel view synthesis quality under sparse training views. Project page: https://***/CoR-GS Copyright © 2024, The Authors. All rights reserved.

关键词： Rendering (computer graphics)

Risk Region-based Prediction Model for the Epidemic Spreading

学校读者我要写书评

暂无评论

Risk Region-based Prediction Model for the Epidemic Spreadin...

International Conference on computer Supported Cooperative Work in Design

作者： Fuwei Li Dianjie Lu Baoxu An Guijuan Zhang School of Information Science and Engineering Shandong Normal University Jinan China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan China State Key Laboratory of High-end Server & Storage Technology Jinan China

Studying the spread process of epidemics in the crowd is a significant topic, which can help to take measures in advance to avoid epidemic spread and reduce the losses of life and property of the general public. However, this is a challenging problem that how to capture the spatio-temporal characteristics of the epidemic spreading. In this paper, we propose a risk region-based prediction model to characterize the spread feature. First, we propose a risk region-based spread model to describe the spatial relationship between individuals by dividing risk regions and assigning different spread characteristics to different risk regions. Second, we develop a spatio-temporal point process-based infection intensity quantification method to describe the variation of epidemic spread characteristics over time in different risk regions in the form of intensity. Then, we propose a long short-term memory (LSTM) based intensity function prediction method to solve the corresponding intensity and predict the spread process by the spatio-temporal point process (STPP). Finally, we implement an epidemic spreading simulation platform to verify our method and visualize the experimental results. The proposed model is expected to provide guidance for predicting the spread of epidemics.

关键词：

Abnormal Crowd Behavior Detection Based on the Fusion of Macro and Micro Features

学校读者我要写书评

暂无评论

Abnormal Crowd Behavior Detection Based on the Fusion of Mac...

International Conference on computer Supported Cooperative Work in Design

作者： Shourong Jiao Dianjie Lu Chuanhua Jia Hong Liu Guijuan Zhang School of Information Science and Engineering Shandong Normal University Jinan China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan China State Key Laboratory of High-end Server & Storage Technology Jinan China

There are two types of research on abnormal crowd behavior detection: micro modeling methods (e.g., crowd pose) and macro modeling methods (e.g., optical flow methods). In crowd evacuation, however, micro methods fail to solve the occlusion problem, while macro methods fail to address issues such as poor real-time performance and non-adaptive motion target detection. To address these issues, we propose a method that combines macro and micro methods to detect abnormal crowd behavior. Firstly, we extract the two-dimensional poses of moving people using the human pose estimation algorithm (OpenPose) and get the corresponding micro motion features. Secondly, we create an optical flow map of the video using the dense optical flow algorithm (Farneback) and a scene activity map by comparing the optical flow sizes between two consecutive frames. Then we derive the entropy change curve, which represents the macro motion features. Finally, we fuse micro and macro motion features and put the fused hybrid feature vectors into our classification network to train our model, and then find the transition time from micro to macro abnormal states in crowd evacuation videos. The experimental findings demonstrate that our proposed method can more accurately and quickly identify the abnormal behaviors that appear in evacuation scenes.

关键词：

Human Observation-Inspired Trajectory Prediction for Autonomous Driving in Mixed-Autonomy Traffic Environments

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liao, Haicheng Liu, Shangqian Li, Yongkang Li, Zhenning Wang, Chengyue Li, Yunjian Li, Shengbo Eben Xu, Chengzhong State Key Laboratory of Internet of Things for Smart City Department of Computer and Information Science University of Macau China Department of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China State Key Laboratory of Internet of Things for Smart City Department of Civil and Environmental Engineering and Computer and Information Science University of Macau China State Key Laboratory of Internet of Things for Smart City Department of Civil and Environmental Engineering University of Macau China Faculty of Innovation Engineering Macau University of Science and Technology China School of Vehicle and Mobility Tsinghua University Beijing China

In the burgeoning field of autonomous vehicles (AVs), trajectory prediction remains a formidable challenge, especially in mixed autonomy environments. Traditional approaches often rely on computational methods such as time-series analysis. Our research diverges significantly by adopting an interdisciplinary approach that integrates principles of human cognition and observational behavior into trajectory prediction models for AVs. We introduce a novel "adaptive visual sector" mechanism that mimics the dynamic allocation of attention human drivers exhibit based on factors like spatial orientation, proximity, and driving speed. Additionally, we develop a "dynamic traffic graph" using Convolutional Neural Networks (CNN) and Graph Attention Networks (GAT) to capture spatio-temporal dependencies among agents. Benchmark tests on the NGSIM, HighD, and MoCAD datasets reveal that our model (GAVA) outperforms state-of-the-art baselines by at least 15.2%, 19.4%, and 12.0%, respectively. Our findings underscore the potential of leveraging human cognition principles to enhance the proficiency and adaptability of trajectory prediction algorithms in AVs. The code for the proposed model is available at our Github. © 2024, CC BY-NC-SA.

关键词： Time series analysis

Deep reinforcement learning-based panic crowd evacuation simulation

学校读者我要写书评

暂无评论

Deep reinforcement learning-based panic crowd evacuation sim...

International Conference on computer Supported Cooperative Work in Design

作者： Baoxu An Guijuan Zhang Fuwei Li Nianyun Song Hong Liu Dianjie Lu School of Information Science and Engineering Shandong Normal University Jinan China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology Jinan China State Key Laboratory of High-end Server & Storage Technology Jinan China

Crowd evacuation simulation can provide guidance for public emergencies or mass casualties which is easy to cause crowding and stampede. Deep reinforcement learning is an effective method for the path planning of crowd evacuation simulation which can reduce the dependence on data and has strong generalization. However, current methods of deep reinforcement learning ignore the consideration of panic emotion which could result in weak authenticity since the panic emotion has a significant impact on crowd evacuation. To address this problem, we propose a novel deep reinforcement learning for the panic crowd evacuation. First, a quadrant based crowd network is constructed to guide the movement of agents according to the change of panic degree in different quadrants. Second, a method of quantifying panic degree is proposed. Through the mean field equation, the dynamic evolution of agent states is realized, and the panic degree in different regions is quantified. Finally, a panic crowd evacuation model based on multi-agent deep deterministic policy gradient (MADDPG) was established. The computated panic degree is introduced into the reward function to recommend a path for multiagent to avoid panic. Experimental results show the effectiveness of our method.

关键词：

SG-GS: Topology-aware Human Avatars with Semantically-guided Gaussian Splatting

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Haoyu Yang, Chen Wang, Hao Zhao, Xingyue Shen, Wei MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University China School of Computer Science Wuhan University China Wuhan National Laboratory for Optoelectronics Huazhong University of Science and Technology China School of Software Engineering Xi’an Jiao Tong University China

Reconstructing photo-realistic and topology-aware animatable human avatars from monocular videos remains challenging in computer vision and graphics. Recently, methods using 3D Gaussians to represent the human body have emerged, offering faster optimization and real-time rendering. However, due to ignoring the crucial role of human body semantic information which represents the explicit topological and intrinsic structure within human body, they fail to achieve fine-detail reconstruction of human avatars. To address this issue, we propose SG-GS, which uses semantics-embedded 3D Gaussians, skeleton-driven rigid deformation, and non-rigid cloth dynamics deformation to create photo-realistic human avatars. We then design a Semantic Human-Body Annotator (SHA) which utilizes SMPL’s semantic prior for efficient body part semantic labeling. The generated labels are used to guide the optimization of semantic attributes of Gaussian. To capture the explicit topological structure of the human body, we employ a 3D network that integrates both topological and geometric associations for human avatar deformation. We further implement three key strategies to enhance the semantic accuracy of 3D Gaussians and rendering quality: semantic projection with 2D regularization, semantic-guided density regularization and semantic-aware regularization with neighborhood consistency. Extensive experiments demonstrate that SG-GS achieves state-of-the-art geometry and appearance reconstruction performance. Our project is at https://***/. Copyright © 2024, The Authors. All rights reserved.

关键词： Topology