检索结果-内蒙古大学图书馆

arXiv 2024年

作者： He, Xilin Hu, Jingyu Lin, Qinliang Luo, Cheng Xie, Weicheng Song, Siyang Khan, Muhammad Haris Shen, Linlin Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China Guangdong Provincial Key Laboratory of Intelligent Information Processing China University of Exeter United Kingdom Mohamed bin Zayed University of Artificial Intelligence United Arab Emirates

Domain generalization methods aim to learn transferable knowledge from source domains that can generalize well to unseen target domains. Recent studies show that neural networks frequently suffer from a simplicity-biased learning behavior which leads to over-reliance on specific frequency sets, namely as frequency shortcuts, instead of semantic information, resulting in poor generalization performance. Despite previous data augmentation techniques successfully enhancing generalization performances, they intend to apply more frequency shortcuts, thereby causing hallucinations of generalization improvement. In this paper, we aim to prevent such learning behavior of applying frequency shortcuts from a data-driven perspective. Given the theoretical justification of models’ biased learning behavior on different spatial frequency components, which is based on the dataset frequency properties, we argue that the learning behavior on various frequency components could be manipulated by changing the dataset statistical structure in the Fourier domain. Intuitively, as frequency shortcuts are hidden in the dominant and highly dependent frequencies of dataset structure, dynamically perturbating the over-reliance frequency components could prevent the application of frequency shortcuts. To this end, we propose two effective data augmentation modules designed to collaboratively and adaptively adjust the frequency characteristic of the dataset, aiming to dynamically influence the learning behavior of the model and ultimately serving as a strategy to mitigate shortcut learning. Code is available at AdvFrequency. © 2024, CC BY.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

3D Feature Tracking via Event Camera

3D Feature Tracking via Event Camera

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Siqi Li Zhikuan Zhou Zhou Xue Yipeng Li Shaoyi Du Yue Gao BNRist THUIBCS School of Software Tsinghua University Li Auto Department of Automation Tsinghua University National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xian Jiaotong University

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

This paper presents the first 3D feature tracking method with the corresponding dataset. Our proposed method takes event streams from stereo event cameras as input to pre-dict 3D trajectories of the target features with high-speed motion. To achieve this, our method leverages a joint framework to predict the 2D feature motion offsets and the 3D feature spatial position simultaneously. A motion compensation module is leveraged to overcome the feature deformation. A patch matching module based on bi-polarity hypergraph modeling is proposed to robustly es-timate the feature spatial position. Meanwhile, we collect the first 3D feature tracking dataset with high-speed moving objects and ground truth 3D feature trajectories at 250 FPS, named E-3DTrack, which can be used as the first high-speed 3D feature tracking benchmark. Our code and dataset could be found at: https://***/lisiqi19971013/E-3DTrack.

关键词： Three-dimensional displays Target tracking Tracking Deformation Benchmark testing Cameras Motion compensation

来源：评论

学校读者我要写书评

暂无评论

MotionTrack: Learning Robust Short-Term and Long-Term Motions for Multi-Object Tracking

MotionTrack: Learning Robust Short-Term and Long-Term Motion...

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Zheng Qin Sanping Zhou Le Wang Jinghai Duan Gang Hua Wei Tang National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University School of Software Engineering Xi'an Jiaotong University Wormpex AI Research University of Illinois at Chicago

The main challenge of Multi-Object Tracking (MOT) lies in maintaining a continuous trajectory for each target. Existing methods often learn reliable motion patterns to match the same target between adjacent frames and discriminative appearance features to re-identify the lost targets after a long period. However, the reliability of motion prediction and the discriminability of appearances can be easily hurt by dense crowds and extreme occlusions in the tracking process. In this paper, we propose a simple yet effective multi-object tracker, i.e., MotionTrack, which learns robust short-term and long-term motions in a unified framework to associate trajectories from a short to long range. For dense crowds, we design a novel Interaction Module to learn interaction-aware motions from short-term trajectories, which can estimate the complex movement of each target. For extreme occlusions, we build a novel Refind Module to learn reliable long-term motions from the target's history trajectory, which can link the interrupted trajectory with its corresponding detection. Our Interaction Module and Refind Module are embedded in the well-known tracking-by-detection paradigm, which can work in tandem to maintain superior performance. Extensive experimental results on MOT17 and MOT20 datasets demonstrate the superiority of our approach in challenging scenarios, and it achieves state-of-the-art performances at various MOT metrics. Code is available at https://***/qwomeng/MotionTrack.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CodeEnhance: A Codebook-Driven Approach for Low-Light Image Enhancement

arXiv

引用

arXiv 2024年

作者： Wu, Xu Hou, XianXu Lai, Zhihui Zhou, Jie Zhang, Ya-Nan Pedrycz, Witold Shen, Linlin The Computer Vision Institute College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen518060 China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen518060 China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society Guangdong Shenzhen518060 China The Department of Electrical & Computer Engineering University of Alberta University of Alberta Canada

Low-light image enhancement (LLIE) aims to improve low-illumination images. However, existing methods face two challenges: (1) uncertainty in restoration from diverse brightness degradations;(2) loss of texture and color information caused by noise suppression and light enhancement. In this paper, we propose a novel enhancement approach, CodeEnhance, by leveraging quantized priors and image refinement to address these challenges. In particular, we reframe LLIE as learning an image-to-code mapping from low-light images to discrete codebook, which has been learned from high-quality images. To enhance this process, a Semantic Embedding Module (SEM) is introduced to integrate semantic information with low-level features, and a Codebook Shift (CS) mechanism, designed to adapt the pre-learned codebook to better suit the distinct characteristics of our low-light dataset. Additionally, we present an Interactive Feature Transformation (IFT) module to refine texture and color information during image reconstruction, allowing for interactive enhancement based on user preferences. Extensive experiments on both real-world and synthetic benchmarks demonstrate that the incorporation of prior knowledge and controllable information transfer significantly enhances LLIE performance in terms of quality and fidelity. The proposed CodeEnhance exhibits superior robustness to various degradations, including uneven illumination, noise, and color distortion. Copyright © 2024, The Authors. All rights reserved.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

HSIDMamba: Exploring Bidirectional State-Space Models for Hyperspectral Denoising

arXiv

引用

arXiv 2024年

作者： Liu, Yang Xiao, Jiahua Guo, Yu Jiang, Peilin Yang, Haiwei Wang, Fei National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Shaanxi Xi’an710049 China School of software Xi’an Jiaotong University Shaanxi Xi’an710049 China

Effectively discerning spatial-spectral dependencies in HSI denoising is crucial, but prevailing methods using convolution or transformers still face computational efficiency limitations. Recently, the emerging Selective State Space Model(Mamba) has risen with its nearly linear computational complexity in processing natural language sequences, which inspired us to explore its potential in handling long spectral sequences. In this paper, we propose HSIDMamba(HSDM), tailored to exploit the linear complexity for effectively capturing spatial-spectral dependencies in HSI denoising. In particular, HSDM comprises multiple Hyperspectral Continuous Scan Blocks, incorporating BCSM(Bidirectional Continuous Scanning Mechanism), scale residual, and spectral attention mechanisms to enhance the capture of long-range and local spatial-spectral information. BCSM strengthens spatial-spectral interactions by linking forward and backward scans and enhancing information from eight directions through SSM, significantly enhancing the perceptual capability of HSDM and improving denoising performance more effectively. Extensive evaluations against HSI denoising benchmarks validate the superior performance of HSDM, achieving state-of-the-art results in performance and surpassing the efficiency of the latest transformer architectures by 30%. Copyright © 2024, The Authors. All rights reserved.

关键词： Computational efficiency

来源：评论

学校读者我要写书评

暂无评论

HCF-Net: Hybrid Coarse-to-Fine Network for Forgery Reconstruction

HCF-Net: Hybrid Coarse-to-Fine Network for Forgery Reconstru...

引用

2021 International Workshop on Safety and Security of Deep Learning, SSDL 2021

作者： Zhuo, Long Tan, Shunquan Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security Shenzhen Institute of Artificial Intelligence and Robotics for Society China College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China

Due to the ubiquity of photo editing software, it is convenient and prevalent to create fake images which may cause terrible misunderstandings. To address this issue, we introduce forgery reconstruction, a novel image reconstruction that automatically reconstructs a forged image into its original appearance. To achieve this, we present a new two-phase system, including a forgery localization phase and a reconstruction phase, where we first localize forged regions and then inpaint the natural texture. The proposed approach, called HCF-Net, is a hybrid two coarse-to-fine networks. Unlike the existing solutions of image reconstruction that relies on the prior knowledge of the areas to be recovered, our proposed system is free from the mask information. To guide the reconstruction, we propose a new strategy for extracting noise features in the forgery localization phase, which leverages the relationships among the image channels and localizes the tampered regions more precisely. For better performance, we equip a dual attention module in the reconstruction phase. Our automatic system is applicable to localize the tampered areas and reconstruct the missing part in a single pass. The extensive experiments demonstrate that our system achieves state-of-the-art performance in forgery localization and generates higher-quality and more flexible results than traditional inpainting methods. © 2021 CEUR-WS. All rights reserved.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

A Semi-Supervised Deep Learning Approach for Cropped Image Detection

SSRN

引用

SSRN 2023年

作者： Hussain, Israr Tan, Shunquan Huang, Jiwu College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China College of Electronic and Information Engineering Shenzhen Key Laboratory of Media Security Shenzhen University Shenzhen518060 China Shenzhen Institute of Artificial Intelligence and Robotics for Society Shenzhen University Shenzhen518000 China

Deep learning algorithms have shown great performance in multimedia forensics applications using supervised learning on large-scale labeled data. However, obtaining a large-scale labeled dataset can be challenging and expensive. Additionally, heavy deep learning models with complex architectures and a large number of parameters require significant hardware resources for training. To address the issues typically associated with image cropping detection, we propose a low-cost semi-supervised deep learning framework that can be trained on a large number of unlabeled image samples. In this framework, a teacher model trained on a small number of labeled image samples is used to sort the recognition confidence of image samples in a large-scale unlabeled dataset. Then, a robust student network is successfully trained using the sorted image samples. We conducted a large number of experiments on a large-scale database to verify the effectiveness of the proposed collaborative training framework in different image cropping detection scenarios. The results show that compared to existing supervised learning detection frameworks, the proposed semi-supervised learning framework can significantly improve the detection performance when labeled image samples are scarce. We further investigated the impact of different teacher/student model pairings, the number of unlabeled samples, the number of self-training iterations, and the proportion of high-confidence samples used in semisupervised learning on the detection performance. © 2023, The Authors. All rights reserved.

关键词： Learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Intelligent design of mechanical metamaterials: a GCNN-based structural genome database approach

引用

National Science Review 2025年第04期 268-281页

作者： Wenyu Hao Zongliang Du Xiuquan Hou Yilin Guo Chang Liu Weisheng Zhang Huajian Gao Xu Guo State Key Laboratory of Structural Analysis Optimization and CAE Software for Industrial EquipmentDepartment of Engineering Mechanics Dalian University of Technology Ningbo Institute of Dalian University of Technology Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Mechano-X Institute Applied Mechanics LaboratoryDepartment of Engineering Mechanics Tsinghua University

The reciprocal mapping between the geometry and properties of a unit cell is crucial for the intelligent and inverse design of advanced materials and structural *** classical homogenization-based numerical methods,this paper presents an efficient and accurate mapping between the geometry and properties of a class of unit cells described by moving morphable components,achieved via a graph convolutional neural *** leads to a structural genome database(SGD) approach for the intelligent design of mechanical *** the SGD approach,metamaterials exhibiting the Hashin-Shtrikman upper bound of bulk modulus,auxetic behavior and the unimodal property have been created,with design efficiency improved by 3-4 orders of ***,transfer learning and a small amount of training data allow the SGD to predict non-local behaviors beyond a unit cell,such as optimized unit cells with critical buckling strength enhanced by nearly 200% and a bandgap metamaterial with a relative bandgap width of 51%.Experimentally validated optimized metamaterials demonstrate auxetic behavior and superior buckling *** proposed SGD approach holds promise for the advanced design of multi-scale and multi-physics systems.

关键词：

来源：评论

学校读者我要写书评

暂无评论

UniUIR: Considering Underwater Image Restoration as An All-in-One Learner

arXiv

引用

arXiv 2025年

作者： Zhang, Xu Zhang, Huan Wang, Guoli Zhang, Qian Zhang, Lefei Du, Bo The Institute of Artificial Intelligence School of Computer Science Wuhan University Wuhan430072 China The Hubei Luojia Laboratory Wuhan China The National Engineering Research Center for Multimedia Software Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan430072 China The School of Information Engineering Guangdong University of Technology Guangzhou510006 China The Horizon Robotics Beijing100083 China

Existing underwater image restoration (UIR) methods generally only handle color distortion or jointly address color and haze issues, but they often overlook the more complex degradations that can occur in underwater scenes. To address this limitation, we propose a Universal Underwater Image Restoration method, termed as UniUIR, considering the complex scenario of real-world underwater mixed distortions as an all-in-one manner. To decouple degradation-specific issues and explore the inter-correlations among various degradations in UIR task, we designed the Mamba Mixture-of-Experts module. This module enables each expert to identify distinct types of degradation and collaboratively extract task-specific priors while maintaining global feature representation based on linear complexity. Building upon this foundation, to enhance degradation representation and address the task conflicts that arise when handling multiple types of degradation, we introduce the spatial-frequency prior generator. This module extracts degradation prior information in both spatial and frequency domains, and adaptively selects the most appropriate task-specific prompts based on image content, thereby improving the accuracy of image restoration. Finally, to more effectively address complex, region-dependent distortions in UIR task, we incorporate depth information derived from a large-scale pre-trained depth prediction model, thereby enabling the network to perceive and leverage depth variations across different image regions to handle localized degradation. Extensive experiments demonstrate that UniUIR can produce more attractive results across qualitative and quantitative comparisons, and shows strong generalization than state-of-the-art methods. Copyright © 2025, The Authors. All rights reserved.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Domain Adaptative Driving Behavior Recognition Through Skeleton-Guided Domain Adversarial Learning

Domain Adaptative Driving Behavior Recognition Through Skele...

引用

International Conference on Intelligent Transportation

作者： Zhiyong Wang Zhiqiang Tian Shaoyi Du School of Software Engineering Xi'an Jiaotong University Xi'an Shaanxi Province P.R. China Institute of Artificial Intelligence and Robotics College of Artificial Intelligence Xi'an Jiaotong University Xi'an Shanxi Province P.R. China

Driving behavior recognition plays an indispensable role in human-centered intelligent transportation systems. However, the diverse range of scenarios and drivers in practical applications poses a significant challenge for existing methods due to their limited domain generalization ability. To improve the cross-domain performance, we propose a domain adaptive driving behavior recognition method that utilizes skeleton-guided behavior representation and employs graph convolution network (GCN)-assisted domain adversarial learning. First, we propose a novel behavior representation by integrating the driver skeleton with the raw image, which effectively combines high-level behavioral patterns and low-level pixel information to enhance domain invariance. Second, we design a GCN-assisted domain adversarial network that utilizes a graph convolutional network to model the relationships between features of different samples, thereby facilitating more robust domain adaption for driving behavior recognition. Our method outperforms other compared methods in the unsupervised domain adaptation (UDA) tasks across the AUC and State Farm datasets. Moreover, the proposed GCN can serve as a plug-and-play technique to enhance existing unsupervised domain adaptation methods, without the need for additional modifications.

关键词：

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：