检索结果-内蒙古大学图书馆

Adaptive Homophily Clustering: Structure Homophily Graph Learning with Adaptive Filter for Hyperspectral image

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Ding, Yao Kang, Weijie Yang, Aitao Zhang, Zhili Zhao, Junyang Feng, Jie Hong, Danfeng Zheng, Qinghe The Key Laboratory of Intelligent Control PLA Rocket Force University of Engineering Xi’an710025 China The Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education of China Xidian University Xi’an710071 China The Aerospace Information Research Institute Chinese Academy of Sciences Beijing100094 China The School of Electronic Electrical and Communication Engineering University of Chinese Academy of Sciences Beijing100049 China

Hyperspectral image (HSI) clustering has been a fundamental but challenging task with zero training labels. Currently, some deep graph clustering methods have been successfully explored for HSI due to their outstanding performance in effective spatial structural information encoding. Nevertheless, insufficient structural information utilization, poor feature presentation ability, and weak graph update capability limit their performance. Thus, in this paper, a homophily structure graph learning with an adaptive filter clustering method (AHSGC) for HSI is proposed. Specifically, homogeneous region generation is first developed for HSI processing and constructing the original graph. Afterward, an adaptive filter graph encoder is designed to adaptively capture the high and low frequency features on the graph for subsequence processing. Then, a graph embedding clustering self-training decoder is developed with KL Divergence, with which the pseudo-label is generated for network training. Meanwhile, homophily-enhanced structure learning is introduced to update the graph according to the clustering task, in which the orient correlation estimation is adopted to estimate the node connection, and graph edge sparsification is designed to adjust the edges in the graph dynamically. Finally, a joint network optimization is introduced to achieve network self-training and update the graph. The K-means is adopted to express the latent features. Extensive experiments and repeated comparative analysis have verified that our AHSGC contains high clustering accuracy, low computational complexity, and strong robustness. The code source will be available at https://***/DY-HYX. Copyright © 2025, The Authors. All rights reserved.

关键词： Adaptive filters

A Deep Learning Filter for Visual Drone Single Object Tracking 16th

学校读者我要写书评

暂无评论

A Deep Learning Filter for Visual Drone Single Object Tracki...

Workshops held at the 16th European Conference on Computer Vision, ECCV 2020

作者： Zhang, Xin Jiao, Licheng Liu, Xu Li, Xiaotong Zhang, Wenhua Zhu, Hao Zhang, Jie Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education School of Artificial Intelligence Xidian University Xi’an China

ISBN: (纸本)9783030668228

Object tracking is one of the most important topics in computer vision. In visual drone tracking, it is an extremely challenging due to various factors, such as camera motion, partial occlusion, and full occlusion. In this paper, we propose a deep learning filter method to relieve the above problems, which is to obtain a priori position of the object at the subsequent frame and predict its trajectory to follow up the object during occlusion. Our tracker adopts the geometric transformation of the surrounding of the object to prevent the bounding box of the object lost, and it uses context information to integrate its motion trend thereby tracking the object successfully when it reappears. Experiments on the VisDrone-SOT2018 test dataset and the VisDrone-SOT2020 val dataset illustrate the effectiveness of the proposed approach. © 2020, Springer Nature Switzerland AG.

关键词： Drones

Prototype-based classifier learning for long-tailed visual recognition

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2022年第6期65卷 62-76页

作者： Xiu-Shen WEI Shu-Lin XU Hao CHEN Liang XIAO Yuxin PENG School of Computer Science and Engineering Nanjing University of Science and Technology State Key Laboratory of Integrated Services Networks Xidian University Key Lab of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education and Jiangsu Key Lab of Image and Video Understanding for Social Security Wangxuan Institute of Computer Technology Peking University

In this paper, we tackle the long-tailed visual recognition problem from the categorical prototype perspective by proposing a prototype-based classifier learning(PCL) method. Specifically, thanks to the generalization ability and robustness, categorical prototypes reveal their advantages of representing the category semantics. Coupled with their class-balance characteristic, categorical prototypes also show potential for handling data imbalance. In our PCL, we propose to generate the categorical classifiers based on the prototypes by performing a learnable mapping function. To further alleviate the impact of imbalance on classifier generation, two kinds of classifier calibration approaches are designed from both prototype-level and example-level aspects. Extensive experiments on five benchmark datasets, including the large-scale iNaturalist, Places-LT,and imageNet-LT, justify that the proposed PCL can outperform state-of-the-arts. Furthermore, validation experiments can demonstrate the effectiveness of tailored designs in PCL for long-tailed problems.

关键词： long-tailed distribution categorical prototype classifier generation classifier calibration class imbalance

Diff-PCR: Diffusion-Based Correspondence Searching in Doubly Stochastic Matrix Space for Point Cloud Registration

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Qianliang Jiang, Haobo Ding, Yaqing Luo, Lei Xie, Jin Yang, Jian PCA Lab Key Lab of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education Jiangsu Key Lab of Image and Video Understanding for Social Security School of Computer Science and Engineering Nanjing University of Science and Technology China

Efficiently finding optimal correspondences between point clouds is crucial for solving both rigid and nonrigid point cloud registration problems. Existing methods often rely on geometric or semantic feature embedding to establish correspondences and estimate transformations or flow fields. Recently, state-of-the-art methods have employed RAFT-like iterative updates to refine the solution. However, these methods have certain limitations. Firstly, their iterative refinement design lacks transparency, and their iterative updates follow a fixed path during the refinement process, which can lead to suboptimal results. Secondly, these methods overlook the importance of refining or optimizing correspondences (or matching matrices) as a precursor to solving transformations or flow fields. They typically compute candidate correspondences based on distances in the point feature space. However, they only project the candidate matching matrix into some matrix space once with Sinkhorn or dual softmax operations to obtain final correspondences. This one-shot projected matching matrix may be far from the globally optimal one, and these approaches do not consider the distribution of the target matching matrix. In this paper, we propose a novel approach that exploits the Denoising Diffusion Model to predict a searching gradient for the optimal matching matrix within the Doubly Stochastic Matrix Space. Our method incorporates the diffusion model to learn a denoising gradient direction. During the reverse denoising process, our method iteratively searches for better solutions along this denoising gradient, which points towards the maximum likelihood direction of the target matching matrix. To improve efficiency, we employ a lightweight design for our denoising module and utilize the sampling strategy in the Denoising Diffusion Implicit Model (DDIM [27]). Experimental evaluations on the 3DMatch/3DLoMatch and 4DMatch/4DLoMatch datasets demonstrate the effectiveness of our newly designe

关键词： Flow fields

Sparse Prior Is Not All You Need: When Differential Directionality Meets Saliency Coherence for Infrared Small Target Detection

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Fei Fu, Maixia Qian, Yulei Yang, Jian Dai, Yimian Ministry of Education Henan Key Laboratory of Grain Photoelectric Detection and Control Henan University of Technology Zhengzhou China Nanjing Marine Radar Institute Nanjing China PCA Lab Key Lab of Intelligent Perception and Systems for High-Dimensional Information of Ministry of Education Jiangsu Key Lab of Image and Video Understanding for Social Security School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing China

Infrared small target detection is crucial for the efficacy of infrared search and tracking systems. Current tensor decomposition methods emphasize representing small targets with sparsity but struggle to separate targets from complex backgrounds due to insufficient use of intrinsic directional information and reduced target visibility during decomposition. To address these challenges, this study introduces a Sparse Differential Directionality prior (SDD) framework. SDD leverages the distinct directional characteristics of targets to differentiate them from the background, applying mixed sparse constraints on the differential directional images and continuity difference matrix of the temporal component, both derived from Tucker decomposition. We further enhance target detectability with a saliency coherence strategy that intensifies target contrast against the background during hierarchical decomposition. A Proximal Alternating Minimization-based (PAM) algorithm efficiently solves our proposed model. Experimental results on several real-world datasets validate our method’s effectiveness, outperforming ten state-of-the-art methods in target detection and clutter suppression. Our code is available at https://***/GrokCV/SDD. Copyright © 2024, The Authors. All rights reserved.

关键词： Tensors

Direct Distillation between Different Domains

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tang, Jialiang Chen, Shuo Niu, Gang Zhu, Hongyuan Zhou, Joey Tianyi Gong, Chen Sugiyama, Masashi School of Computer Science and Engineering Nanjing University of Science and Technology China Key Laboratory of Intelligent Perception and Systems for High-Dimensional Information Ministry of Education China Jiangsu Key Laboratory of Image and Video Understanding for Social Security China Singapore Center for Advanced Intelligence Project RIKEN Japan The Graduate School of Frontier Sciences The University of Tokyo Japan

Knowledge Distillation (KD) aims to learn a compact student network using knowledge from a large pre-trained teacher network, where both networks are trained on data from the same distribution. However, in practical applications, the student network may be required to perform in a new scenario (i.e., the target domain), which usually exhibits significant differences from the known scenario of the teacher network (i.e., the source domain). The traditional domain adaptation techniques can be integrated with KD in a two-stage process to bridge the domain gap, but the ultimate reliability of two-stage approaches tends to be limited due to the high computational consumption and the additional errors accumulated from both stages. To solve this problem, we propose a new one-stage method dubbed "Direct Distillation between Different Domains" (4Ds). We first design a learnable adapter based on the Fourier transform to separate the domain-invariant knowledge from the domain-specific knowledge. Then, we build a fusion-activation mechanism to transfer the valuable domain-invariant knowledge to the student network, while simultaneously encouraging the adapter within the teacher network to learn the domain-specific knowledge of the target data. As a result, the teacher network can effectively transfer categorical knowledge that aligns with the target domain of the student network. Intensive experiments on various benchmark datasets demonstrate that our proposed 4Ds method successfully produces reliable student networks and outperforms state-of-the-art approaches. Copyright © 2024, The Authors. All rights reserved.

关键词： Distillation

学校读者我要写书评

暂无评论

Triple Point Masking

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Liu, Jiaming Kong, Linghe Wu, Yue Gong, Maoguo Li, Hao Miao, Qiguang Ma, Wenping Qin, Can The Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai200240 China The School of Computer Science and Technology Key Laboratory of Collaborative Intelligence Systems Ministry of Education Xidian University Xi’an710071 China The School of Electronic Engineering Key Laboratory of Collaborative Intelligence Systems Ministry of Education Xidian University Xi’an710071 China The College of Mathematics Science Inner Mongolia Normal University Hohhot010022 China The School of Artificial Intelligence Key Laboratory of Intelligent Perception and Image Understanding Ministry of Education Xidian University Xi’an710071 China The Department of Electrical and Computer Engineering Northeastern University BostonMA02115 United States

Existing 3D mask learning methods encounter performance bottlenecks under limited data, and our objective is to overcome this limitation. In this paper, we introduce a triple point masking scheme, named TPM, which serves as a scalable plug- and-play framework for MAE pre-training to achieve multi-mask learning for 3D point clouds. Specifically, we augment the baseline methods with two additional mask choices (i.e., medium mask and low mask) as our core insight is that the recovery process of an object can manifest in diverse ways. Previous high-masking schemes focus on capturing the global representation information but lack fine-grained recovery capabilities, so that the generated pre-training weights tend to play a limited role in the fine-tuning process. With the support of the proposed TPM, current methods can exhibit more flexible and accurate completion capabilities, enabling the potential autoencoder in the pre-training stage to consider multiple representations of a single 3D point cloud object. In addition, during the fine-tuning stage, an SVM-guided weight selection module is proposed to fill the encoder parameters for downstream networks with the optimal weight, maximizing linear accuracy and facilitating the acquisition of intricate representations for new objects. Extensive experimental results and theoretical analysis show that five baselines equipped with the proposed TPM achieve comprehensive performance improvements on various downstream tasks. © 2025 IEEE. All rights reserved.

关键词： 3D modeling

Towards Data-Centric Long-Tailed image Recognition

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Ma, Yanbiao Jiao, Licheng Liu, Fang Li, Lingling Ma, Wenping Liu, Xu Chen, Puhua Yang, Shuyuan aKey Laboratory of Intelligent Perception and Image Understanding of Ministry of Education Xi’an 710071 China bInternational Research Center for Intelligent Perception and Computation Xi’an 710071 China cJoint International Research Center for brain-like perception and cognition Xi’an 710071 China dCollaborative Innovation Center of Quantum Information of Shaanxi Province Xi’an 710071 China eSchool of Artificial Intelligence Xidian University Xi’an 710071 China

In the context of the long-tailed scenario, models exhibit a strong demand for high-quality data. Data-centric approaches aim to enhance both the quantity and quality of data to improve model performance. Among these approaches, information augmentation has been progressively introduced as a crucial category. It achieves a balance in model performance by augmenting the richness and quantity of samples in the tail classes. However, there is currently a lack of research into the underlying mechanisms explaining the effectiveness of information augmentation methods. Consequently, the utilization of information augmentation in long-tail recognition tasks relies heavily on empirical and intricate *** work makes two primary contributions. Firstly, we approach the problem from the perspectives of feature diversity and distribution shift, introducing the concept of Feature Diversity Gain (FDG) to elucidate why information augmentation is effective. We find that the performance of information augmentation can be explained by FDG, and its performance peaks when FDG achieves an appropriate balance. Experimental results demonstrate that by using FDG to select augmented data, we can further enhance model performance without the need for any modifications to the model's architecture. Thus, data-centric approaches hold significant potential in the field of long-tail recognition, beyond the development of new model structures. Furthermore, we systematically introduce the core components and fundamental tasks of a data-centric long-tail learning framework for the first time. The core components are used to guide the implementation and deployment of the system, and the corresponding basic tasks provide rich research directions for the future. © 2024, The Authors. All rights reserved.

关键词： image recognition

EEG-based Emotion Style Transfer Network for Cross-dataset Emotion Recognition

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhou, Yijin Li, Fu Li, Yang Ji, Youshuo Zhang, Lijian Chen, Yuanfang Zheng, Wenming Shi, Guangming Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education The School of Artificial Intelligence Xidian University Xi’an China Beijing Institute of Mechanical Equipment Beijing China School of Biological Sciences and Medical Engineering Southeast University Jiangsu Nanjing China

As the key to realizing aBCIs, EEG emotion recognition has been widely studied by many researchers. Previous methods have performed well for intra-subject EEG emotion recognition. However, the style mismatch between source domain (training data) and target domain (test data) EEG samples caused by huge inter-domain differences is still a critical problem for EEG emotion recognition. To solve the problem of cross-dataset EEG emotion recognition, in this paper, we propose an EEG-based Emotion Style Transfer Network (E2STN) to obtain EEG representations that contain the content information of source domain and the style information of target domain, which is called stylized emotional EEG representations. The representations are helpful for cross-dataset discriminative prediction. Concretely, E2STN consists of three modules, i.e., transfer module, transfer evaluation module, and discriminative prediction module. The transfer module encodes the domain-specific information of source and target domains and then re-constructs the source domain’s emotional pattern and the target domain’s statistical characteristics into the new stylized EEG representations. In this process, the transfer evaluation module is adopted to constrain the generated representations that can more precisely fuse two kinds of complementary information from source and target domains and avoid distorting. Finally, the generated stylized EEG representations are fed into the discriminative prediction module for final classification. Extensive experiments show that the E2STN can achieve the state-of-the-art performance on cross-dataset EEG emotion recognition tasks. Copyright © 2023, The Authors. All rights reserved.

关键词： Emotion Recognition