检索结果-内蒙古大学图书馆

GDPS: A general distillation architecture for end-to-end person search

Journal of Visual Communication and Image Representation 2025年 110卷

作者： Fu, Shichang Lu, Tao Wang, Jiaming Gu, Yu Cai, Jiayi Jiang, Kui Wuhan Institute of Technology Hubei Key Laboratory of Intelligent Robot Wuhan China Faculty of Computing Harbin Institute of Technology Harbin China

Existing knowledge distillation methods for person search tasks handle detection and re-identification (re-id) tasks separately, which may lead to feature conflicts between the two subtasks. On the one hand, by distilling only the detection task, the network will focus more on the common features of pedestrians, which may affect the performance of re-id. On the other hand, by distilling only the re-id task, the network will be more inclined to focus on the personality characteristics of pedestrians, which may harm the detection performance. To solve this problem, we propose a novel distillation method for person search tasks, treating person search as a single task and distilling different tasks in a unified framework, which is called General Distillation for Person Search (GDPS). Specifically, we optimize the general features of detection and re-id by distilling feature-based knowledge, aiming for accurate localization of individuals. In addition, we focus on the re-id task and perform relationship-based and response-based knowledge distillation to obtain more discriminative person features. Finally, we integrate feature-based, relation-based and response-based knowledge into a general framework to achieve simultaneous distillation of two sub-tasks, which can be readily applied to various end-to-end person search methods. Extensive experiments demonstrate the effectiveness of GDPS across different one-step person search methods. Specifically, AlginPS with ResNet-50 achieves 94.1% in mAP with GDPS on the CUHK-SYSU dataset, which surpasses the baseline 93.1% by 1.0%, and is even better than the ResNet-50 DCN-based teacher model with 94.0% mAP. © 2025 Elsevier Inc.

关键词： Knowledge acquisition

来源：评论

学校读者我要写书评

暂无评论

Graph-geometric message passing via a graph convolution transformer for FKP regression

引用

Science China(Information Sciences) 2024年第12期67卷 176-190页

作者： Huizhi ZHU Wenxia XU Jian HUANG Baocheng YU Hubei Key Laboratory of Intelligent Robot Wuhan Institute of Technology School of Artificial Intelligence and Automation Huazhong University of Science and Technology

In this paper, the forward kinematics problem(FKP) of the Gough-Stewart platform(GSP) with six degrees of freedom(6 DoFs) is estimated via deep learning. We propose a graph convolution transformer model by systematically analyzing some challenges encountered with using deep learning regression on largescale data. We attempt to leverage the graph-geometric message as input and singular value decomposition(SVD) orthogonalization for SO(3) manifold learning. This study is the first in which a robot with a sophisticated closed-loop mechanism is described by a graph structure and a specific deep learning model is proposed to solve the FKP of the GSP. Qualitative and quantitative experiments on our dataset demonstrate that our model is feasible and superior to other methods. Our method can guarantee error percentages of translation and rotation less than 1 mm and 1° of 81.9% and 96.7%, respectively.

关键词： deep learning graph-structured learning graph convolution transformer forward kinematics problem Gough-Stewart platform

来源：评论

学校读者我要写书评

暂无评论

THE PIT BOSS

引用

IEEE SPECTRUM 2017年第5期54卷 32-50页

作者： Perry, Tekla S. School of Computer Science and Engineering Hubei Key Laboratory of intelligent robot Wuhan Institute of Technology Wuhan China

Remember vinyl records? More specifically, do you remember the way vinyl records skip when they're dusty or scratched? Let me assume you're old enough to recall that annoyance, or perhaps you've experienced that vintage technology more recently. Now think back to when you got your first CD. Small and shiny, packing 74 minutes of music, it seemed magical, even more magical when you noticed that you could treat a disc pretty badly before physical damage affected the way it played. A lot of different kinds of engineering, of course, went into figuring out how to put music on a CD and play it back so reliably. There's hardware, including a laser, optics to focus it, and mechanical systems to move the laser and turn the disc. And there’s software-including pulse-code modulation, which turns regular samples of an analog signal into bits, and error-correcting codes, which make sure those bits don't get corrupted.

关键词： Encoding Companies Servomotors Patents Logic gates DVD Laser theory

来源：评论

学校读者我要写书评

暂无评论

TransAnno-Net: A Deep Learning Framework for Accurate Cell Type Annotation of Mouse Lung Tissue Using Self-supervised Pretraining

引用

Computer Methods and Programs in Biomedicine 2025年 267卷 108809-108809页

作者： Zhang, Qing Wu, Xiaoxiao Li, Xiang Ma, Wei Wu, Tongquan Li, Liuyue Hu, Fan Xie, Yicheng Wu, Xinglong School of Computer Science and Engineering Hubei Key Laboratory of Intelligent Robot Wuhan Institute of Technology Hubei Wuhan430205 China School of Mechanical Engineering Hefei University of Technology Hefei China Faculty of Applied Science and Engineering University of Toronto Canada UNISKIN Research Institute on Skin Aging Inertia Shanghai Biotechnology Co. Ltd Shanghai China DermaHealth Shanghai Biotechnology Co. Ltd. Shanghai China Department of Dermatology Children's Hospital Zhejiang University School of Medicine National Clinical Research Center for Child Health Zhejiang Province Hangzhou310052 China

Background: Single-cell RNA sequencing (scRNA-seq) has become a significant tool for addressing complex issuess in the field of biology. In the context of scRNA-seq analysis, it is imperative to accurately determine the type of each cell. However, conventional supervised or semi-supervised methodologies are contingent on expert labels and incur substantial labeling costs, In contrast self-supervised pre-training strategies leverage unlabeled data during the pre-training phase and utilise a limited amount of labeled data in the fine-tuning phase, thereby greatly reducing labor costs. Furthermore, the fine-tuning does not need to learn the feature representations from scratch, enhancing the efficiency and transferability of the model. Methods: The proposed methodology is outlined below. The deep learning framework, TransAnno-Net, is based on transfer learning and a Transformer architecture. It has been designed for efficient and accurate cell type annotations in large-scale scRNA-seq datasets of mouse lung organs. Specifically, TransAnno-Net is pre-trained on the scRNA-seq lung data of approximately 100,000 cells to acquire gene-gene similarities via self-supervised learning. It is then migrated to a relatively small number of datasets to fine-tune specific cell type annotation tasks. To address the issue of imbalance in cell types commonly observed in scRNA-seq data, we applied a random oversampling technique is applied to the fine-tuned dataset. This is done to mitigate the impact of distributional imbalance on the annotation outcomes. Results: The experimental findings demonstrate that TransAnno-Net exhibits superior performance with an AUC of 0.979, 0.901, and 0.982, respectively, on three mouse lung datasets, outperforming eight state-of-the-art (SOTA) methods. In addition, TransAnno-Net demonstrates robust performance on cross-organ, cross-platform datasets, and is competitive with the fully supervised learning-based method. Conclusion: The TransAnno-Net method

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Distributed Source Localization under Anchor Position Uncertainty

引用

Chinese Journal of Electronics 2014年第1期23卷 93-96页

作者： LIU Hongqing WU Yuntao Department of Electronic and Computer Engineering National University of Singapore Key Laboratory of Intelligent Robot in Hubei Province Wuhan Institute of Technology

In this work, a distributed source positioning approach is developed based on Alternating direction method of multipliers(ADMM). First, a centralized positioning method is developed under case of the anchor uncertainty. And then, the method is realized in a distributed way using ADMM. Simulation results show that the centralized one is robust to the anchor errors and distributed one has similar performance as the centralized one.

关键词： Target positioning Anchor uncertainty Distributed Computation

来源：评论

学校读者我要写书评

暂无评论

COMPACT COMPOSITION OPERATORS ON GENERALIZED LIPSCHITZ SPACES

引用

Acta Mathematica Scientia 2011年第4期31卷 1347-1356页

作者：戴济能欧阳才衡 School of Science Hubei Province Key Laboratory of Intelligent Robot Wuhan Institute of Technology Wuhan Institute of Physics and Mathematics the Chinese Academy of Science

In this paper, boundedness and compactness of the composition operator on the generalized Lipschitz spaces Λα （α 〉 1） of holomorphic functions in the unit disk are characterized.

关键词： composition operator Lipschitz space Bloch-type space angular derivative

来源：评论

学校读者我要写书评

暂无评论

Context-Aware Network for Person Search 4

Context-Aware Network for Person Search

引用

4th International Conference on robotics and Computer Vision, ICRCV 2022

作者： Gu, Yu Lu, Tao Wuhan Institute of Technology Hubei Key Laboratory of Intelligent Robot Wuhan Hubei China

ISBN: (数字)9781665481700

ISBN: (纸本)9781665481700

The key to effective person search is aiming to localize the pedestrians and obtain the discriminative embeddings representation for person ReID from numerous surveillance scene images. And the existing one-step anchor-free methods can achieve a trade-off between speed and accuracy, but it can not fully exploit the contextual feature information of search context, resulting in undesirable localization. To alleviate this issue, we propose a Context-Aware Network for Person Search (CANPS) to delve into the high-level contextual information. In CANPS, firstly, context encoder is proposed to bridge the gap between the feature maps, achieved by distributing rich contextual information to prediction head layers. Second, we design the malleable center sampling strategy to reasonably expose sample region and focus on the centroid feature representations. What's more, we design above components in a trainable bag-of-freebies manner, so that real-time person search can greatly improve the accuracy without increasing extra inference cost. Extensive experiments show that the approach we proposed can outperform current state-of-the-art methods in public CUHK-SYSU datasets. © 2022 IEEE.

关键词： Economic and social effects

来源：评论

学校读者我要写书评

暂无评论

Single Image Deraining via Two Stages: Bilateral Learning and Feature Decoupling 4

Single Image Deraining via Two Stages: Bilateral Learning an...

引用

4th International Conference on robotics and Computer Vision, ICRCV 2022

作者： Jiang, Rui Liu, Wei Wuhan Institute of Technology Hubei Key Laboratory of Intelligent Robot Hubei Wuhan China

ISBN: (数字)9781665481700

ISBN: (纸本)9781665481700

Single image deraining is an important technique in the field of computer vision. Existing deraining methods remove rain streaks from an image by recovering a clear image directly from a rainy image or by learning the residual components. However, there is a mixed correlation between rain streaks and background information, and it is difficult to remove rain streaks directly from rain images. In this work, to improve the performance and efficiency of image deraining, we propose a novel two-stage rain removal network. In the first stage, we first extract rain-related features in low-resolution paths and map them into bilateral space. Then, affine curves are fitted in the bilateral space. Finally, a slicing operation is performed on the data-dependent lookups in the bilateral grid to separate the rain traces from the background information with less computational effort. Then, in the second stage we design a progressive separation module (PSM) to further decouple the mixing correlation between the background and rain-scar layers in a progressive coupled representation learning manner. Tested on several synthetic rainfall datasets as well as real natural scene images, qualitative and quantitative results show that our proposed method achieves better performance with lower computational resources. © 2022 IEEE.

关键词： Rain

来源：评论

学校读者我要写书评

暂无评论

Single Image Defogging via Recurrent Bilateral Learning 4

Single Image Defogging via Recurrent Bilateral Learning

引用

4th International Conference on robotics and Computer Vision, ICRCV 2022

作者： Chen, Cheng Liu, Wei Lu, Tao Wuhan Institute of Technology Hubei Key Laboratory of Intelligent Robot Hubei Wuhan China

ISBN: (数字)9781665481700

ISBN: (纸本)9781665481700

Image defogging is an important computer vision topic for a long time. Current end-to-end defogging methods based on convolutional neural networks (CNNs) have achieved significant success. However, the inherent ambiguity between scene albedo and depth in foggy degraded images makes foggy image restoration still challenging. In this paper, we introduce a novel fog removal network based on recurrent bilateral learning (RBL) that generates detailed scene depth features and then fully leverages the scene depth to resolve this ambiguity. This network is composed of multiple sets of bilateral grid modules (BGM). In the BGM, we first extract haze-related features in low-resolution streams. Then, the scene depth information is separated from a trainable bilateral grid. Finally, it performs slicing operations for data-dependent lookup in the bilateral grid to reconstruct scene depth with high-frequency features. Moreover, we embed Conv-LSTM in the BGM to better correlate context-relevant information in recurrent structure. Experimental results demonstrate that the proposed method performs favorably against the state-of-the-art algorithms. © 2022 IEEE.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Automatic face replacement in photographs based on active shape models

Automatic face replacement in photographs based on active sh...

引用

2009 2nd Asia-Pacific Conference on Computational Intelligence and Industrial Applications, PACIIA 2009

作者： Min, Feng Lu, Tongwei Zhang, Yanduo Hubei Key Laboratory of Intelligent Robot Wuhan Institute of Technology China

ISBN: (纸本)9781424446070

In this paper, we present an automatic face replacement approach in photographs based on Active Shape Models (ASM). Our replacement algorithm has three main modules: face alignment, face morph, and seamless blending. First, given an input and target image, we automatically implement face alignment using ASM Second, we fit a morphable 2D model to the input face, and warp its shape to match the target face shape. Last, we adjust the lighting and color of the target face image to match those of the input face image, and seamlessly blend the target face image into the input image. Our approach is 2D model-based, fully automatic, and generates natural results. The improved results demonstrate the effectiveness of our approach. ©2009 IEEE.

关键词： Photography

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：