检索结果-内蒙古大学图书馆

Conference on computer vision and Pattern Recognition (CVPR)

作者： Qucheng Peng Ce Zheng Chen Chen Center for Research in Computer Vision University of Central Florida

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

3D human pose data collected in controlled laboratory settings present challenges for pose estimators that generalize across diverse scenarios. To address this, domain generalization is employed. Current methodologies in do-main generalization for 3D human pose estimation typically utilize adversarial training to generate synthetic poses for training. Nonetheless, these approaches exhibit several limitations. First, the lack of prior information about the target domain complicates the application of suitable augmentation through a single pose augmentor, affecting generalization on target domains. Moreover, adversarial training's discriminator tends to enforce similarity between source and synthesized poses, impeding the exploration of out-of-source distributions. Furthermore, the pose estimator's op-timization is not exposed to domain shifts, limiting its over-all generalization ability. To address these limitations, we propose a novel frame-work featuring two pose augmentors: the weak and the strong augmentors. Our framework employs differential strategies for generation and discrimination processes, facilitating the preservation of knowledge related to source poses and the exploration of out-of-source distributions without prior information about target poses. Besides, we leverage meta-optimization to simulate domain shifts in the optimization process of the pose estimator, thereby improving its generalization ability. Our proposed approach significantly outperforms existing methods, as demonstrated through comprehensive experiments on various benchmark datasets. Our code will be released at https://***/davidpengucf/DAF-DG.

关键词： Training computer vision Three-dimensional displays Limiting Codes Pose estimation Data augmentation

来源：评论

学校读者我要写书评

暂无评论

Navigating Heterogeneity and Privacy in One-Shot Federated Learning with Diffusion Models

Navigating Heterogeneity and Privacy in One-Shot Federated L...

引用

IEEE Workshop on Applications of computer vision (WACV)

作者： Matías Mendieta Guangyu Sun Chen Chen Center for Research in Computer Vision University of Central Florida USA

ISBN: (数字)9798331510831

ISBN: (纸本)9798331510848

Federated learning (FL) enables multiple clients to train models collectively while preserving data privacy. However, FL faces challenges in terms of communication cost and data heterogeneity. One-shot federated learning has emerged as a solution by reducing communication rounds, improving efficiency, and providing better security against eavesdropping attacks. Nevertheless, data heterogeneity remains a significant challenge, impacting performance. This work explores the effectiveness of diffusion models in oneshot FL, demonstrating their applicability in addressing data heterogeneity and improving FL performance. Additionally, we investigate the utility of our diffusion model approach, FedDiff, compared to other one-shot FL methods under differential privacy (DP). Furthermore, to improve generated sample quality under DP settings, we propose a pragmatic Fourier Magnitude Filtering (FMF) method, enhancing the effectiveness of the generated data for global model training. Code available at https://***/mmendiet/FedDiff.

关键词： Training Federated learning Filtering Navigation Diffusion models Data models Security Faces Pragmatics Eavesdropping

来源：评论

学校读者我要写书评

暂无评论

BC- MRI -SEG: A Breast Cancer MRI Tumor Segmentation Benchmark

BC- MRI -SEG: A Breast Cancer MRI Tumor Segmentation Benchma...

引用

IEEE International Conference on Healthcare Informatics (ICHI)

作者： Anthony Bilic Chen Chen Center for Research in Computer Vision University of Central Florida

ISBN: (数字)9798350383737

ISBN: (纸本)9798350383744

Binary breast cancer tumor segmentation with Mag-netic Resonance Imaging (MRI) data is typically trained and evaluated on private medical data, which makes comparing deep learning approaches difficult. We propose a benchmark (BC-MRI -SEG) for binary breast cancer tumor segmentation based on publicly available MRI datasets. The benchmark consists of four datasets in total, where two datasets are used for supervised training and evaluation, and two are used for zero-shot evaluation. Additionally we compare state-of-the-art (SOTA) approaches on our benchmark and provide an exhaustive list of available public breast cancer MRI datasets. The source code has been made available at https://***_MRI_SEG_Benchmarkl.

关键词： Training Image segmentation Magnetic resonance imaging Source coding Medical services Benchmark testing Breast cancer

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Object Detection using Patch Based Image Classifier and Gradient Importance Map

引用

International Journal of Information Technology (Singapore) 2025年第4期17卷 2407-2416页

作者： Jain, Vanita Pillai, Manu S. Jain, Achin Dubey, Arun Kumar Department of Electronic Science University of Delhi Delhi India Center for Research in Computer Vision University of Central Florida Orlando United States Bharati Vidyapeeth’s College of Engineering New Delhi India

Image classification in computer vision has seen tremendous amount of success in recent years. Deep learning has played a pivotal role in achieving human level performance in many image recognition challenges and benchmarks. Even though, image classification has been so successful, no other closely related domains have taken advantage from the efforts put into development of image classification methods. One such closely related field is of Object Detection. Object detection or localisation is a computer vision problem whose solutions have not been victorious enough to human level performance. Many challenges arise when developing object detection models for newly generated domains, one of which is labelling of datasets. Preparation of dataset is one of the most cumbersome and expensive task to accomplish while developing an object detection model. Although, image classifiers are used as a feature extractor in object detection training regimes, their localisation abilities are barely studied. In this paper, we propose an object detection training regime, that does not rely on bounding box labelled datasets, hence unsupervised in nature, and is solely based on trained image classifiers. We build up on our hypothesis, that, "if an image classifier is able to predict what object is in the input image, then it must have information about where the object is, we just need a mechanism to extract that information from it". Precisely, we divide the input image into patches of same size and employ a parameter restricted convolutional classifier on each patches to predict whether it contains the object or not, we call this our patch-based image classifier (the object here is the prediction of the trained image classifier). The training of the patch-based classifier is not straightforward as there is no true labels for each patches on which we can reduce the binary cross-entropy. Therefore, we propose a loss function weighted by the importance map, which we generate using Grad

关键词： Deep learning Grad-CAM Image classification Object detection

来源：评论

学校读者我要写书评

暂无评论

YOLOv8: A Novel Vehicle Mirror Detection Algorithm with Enhanced Performance using Deep Learning

YOLOv8: A Novel Vehicle Mirror Detection Algorithm with Enha...

引用

Electronics and Renewable Systems (ICEARS), International Conference on

作者： Sundarakrishnan Narayanan Yashas Donthi Mohana Ramakanth Kumar P Sreelakshmi K Computer Science and Engineering Center for Computer Vision Research RV College of Engineering® Bengaluru India Electronics & Telecommunication Engineering Center for Computer Vision Research RV College of Engineering® Bengaluru India

ISBN: (数字)9798331509675

ISBN: (纸本)9798331509682

In recent times, Artificial intelligence (AI) and machine learning (ML) technologies have demonstrated remarkable capabilities in addressing real time needs and challenges across multiple domains and fields. In the world of deep learning, various Neural Network architectures like DNN, CNN, and RNN have been instrumental in advancing solutions to complex problems spanning multiple industries, business environments and law enforcement use cases. This paper implemented a special type of CNN, YOLOv8(You Only Look Once), a SOTA architecture to detect the presence of rear-view mirrors in two-wheelers for traffic surveillance and law enforcement applications. The proposed research introduces a novel idea in the realm of road safety by presenting the first ever paper on vehicle mirror detection. This approach sheds light on an often-overlooked aspect, contributing significantly to the enhancement of road safety measures. A custom dataset of 1246 training images, 155 test images and 157 validation images has been considered. This dataset has been used to train and test a binary classification model with the label "mirror". Images with the label "mirror" indicate the presence of one or both rear-view mirrors on two-wheelers. On the other hand, images with no mirrors are treated as "background" images. The model achieved a maximum precision of 93.8%, recall of 87.9%, mAP50 of 94.1% and F1-score of 90.9%. These results stand as evidence of the robustness and proficiency of the model, ensuring that it can be potentially used in real world use cases, specifically for law enforcement applications.

关键词： Deep learning YOLO Training Law enforcement Traffic control Video surveillance Road safety Robustness Mirrors Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Ultra-large mode area multi-core orbital angular momentum transmission fiber designed by neural network and optimization algorithms

引用

Optoelectronics Letters 2023年第12期19卷 744-751页

作者： GU Zhiwei HUANG Wei ZHANG Ran FAN Junjie SONG Binbin Key Laboratory of Computer Vision and Systems(Ministry of Education) School of Computer Science and EngineeringTianjin University of TechnologyTianjin 300384China Engineering Research Center of Learning-Based Intelligent System(Ministry of Education) Tianjin University of TechnologyTianjin 300384China

A large mode area multi-core orbital angular momentum(OAM)transmission fiber is designed and optimized by neural network and optimization *** neural network model has been established first to predict the optical properties of multi-core OAM transmission fibers with high accuracy and speed,including mode area,nonlinear coefficient,purity,dispersion,and effective index *** the trained neural network model is combined with different particle swarm optimization(PSO)algorithms for automatic iterative optimization of multi-core structures *** to the structural advantages of multi-core fiber and the automatic optimization process,we designed a number of multi-core structures with high OAM mode purity(>95%)and ultra-large mode area(>3000µm^(2)),which is larger by more than an order of magnitude compared to the conventional ring-core OAM transmission fibers.

关键词： network fiber OAM

来源：评论

学校读者我要写书评

暂无评论

Segmentation of Head and Neck Tumors Using Dual PET/CT Imaging:Comparative Analysis of 2D,2.5D,and 3D Approaches Using UNet Transformer

引用

computer Modeling in Engineering & Sciences 2024年第12期141卷 2351-2373页

作者： Mohammed A.Mahdi Shahanawaj Ahamad Sawsan A.Saad Alaa Dafhalla Alawi Alqushaibi Rizwan Qureshi Information and Computer Science Department College of Computer Science and EngineeringUniversity of Ha’ilHa’il55476Saudi Arabia Software Engineering Department College of Computer Science and EngineeringUniversity of Ha’ilHa’il55476Saudi Arabia Computer Engineering Department College of Computer Science and EngineeringUniversity of Ha’ilHa’il55476Saudi Arabia Department of Computer and Information Sciences Universiti Teknologi PetronasSeri Iskandar32610Malaysia Center for Research in Computer Vision(CRCV) University of Central FloridaOrlandoFL 32816USA

The segmentation of head and neck(H&N)tumors in dual Positron Emission Tomography/Computed Tomogra-phy(PET/CT)imaging is a critical task in medical imaging,providing essential information for diagnosis,treatment planning,and outcome *** by the need for more accurate and robust segmentation methods,this study addresses key research gaps in the application of deep learning techniques to multimodal medical ***,it investigates the limitations of existing 2D and 3D models in capturing complex tumor structures and proposes an innovative 2.5D UNet Transformer model as a *** primary research questions guiding this study are:(1)How can the integration of convolutional neural networks(CNNs)and transformer networks enhance segmentation accuracy in dual PET/CT imaging?(2)What are the comparative advantages of 2D,2.5D,and 3D model configurations in this context?To answer these questions,we aimed to develop and evaluate advanced deep-learning models that leverage the strengths of both CNNs and *** proposed methodology involved a comprehensive preprocessing pipeline,including normalization,contrast enhancement,and resampling,followed by segmentation using 2D,2.5D,and 3D UNet Transformer *** models were trained and tested on three diverse datasets:HeckTor2022,AutoPET2023,and *** was assessed using metrics such as Dice Similarity Coefficient,Jaccard Index,Average Surface Distance(ASD),and Relative Absolute Volume Difference(RAVD).The findings demonstrate that the 2.5D UNet Transformer model consistently outperformed the 2D and 3D models across most metrics,achieving the highest Dice and Jaccard values,indicating superior segmentation *** instance,on the HeckTor2022 dataset,the 2.5D model achieved a Dice score of 81.777 and a Jaccard index of 0.705,surpassing other model *** 3D model showed strong boundary delineation performance but exhibited variability across datasets,while the

关键词： PET/CT imaging tumor segmentation weighted fusion transformer multi-modal imaging deep learning neural networks clinical oncology

来源：评论

学校读者我要写书评

暂无评论

Stable Mean Teacher for Semi-supervised Video Action Detection 39

Stable Mean Teacher for Semi-supervised Video Action Detecti...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Kumar, Akash Mitra, Sirshapan Rawat, Yogesh Singh Center for Research in Computer Vision University of Central Florida United States

ISBN: (纸本)157735897X

In this work, we focus on semi-supervised learning for video action detection. Video action detection requires spatiotemporal localization in addition to classification and a limited amount of labels makes the model prone to unreliable predictions. We present Stable Mean Teacher, a simple end-to-end student-teacher based framework that benefits from improved and temporally consistent pseudo labels. It relies on a novel ErrOr Recovery (EoR) module which learns from students’ mistakes on labeled samples and transfers this to the teacher to improve pseudo-labels for unlabeled samples. Moreover, existing spatio-temporal losses does not take temporal coherency into account and are prone to temporal inconsistencies. To overcome this, we present Difference of Pixels (DoP), a simple and novel constraint focused on temporal consistency which leads to coherent temporal detections. We evaluate our approach on four different spatio-temporal detection benchmarks, UCF101-24, JHMDB21, AVA, and Youtube-VOS. Our approach outperforms the supervised baselines for action detection by an average margin of 23.5% on UCF101-24, 16% on JHMDB21, and, 3.3% on AVA. Using merely 10% and 20% of data, it provides a competitive performance compared to the supervised baseline trained on 100% annotations on UCF101-24 and JHMDB21 respectively. We further evaluate its effectiveness on AVA for scaling to large-scale datasets and Youtube-VOS for video object segmentation demonstrating its generalization capability to other tasks in the video domain. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Semi-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Multi-stage Conditional GAN Architectures for Person-Image Generation 1

引用

1st International Conference and 2nd International Conference on Deep Learning Theory and Applications, DeLTA 2020 and DeLTA 2021

作者： Kurupathi, Sheela Raju Dumpala, Veeru Stricker, Didier Department of Computer Science Technical University of Kaiserslautern Kaiserslautern Germany Augmented Vision German Research Center for Artificial Intelligence Kaiserslautern Germany

ISBN: (数字)9783031373206

ISBN: (纸本)9783031373190

Generating realistic human images has been of great value in recent times due to their varied application in Robotics, computer Graphics, Movie Making, and Games. Advancements in Artificial Intelligence (AI) and Machine learning (ML) lead to the rapid growth of integrating every aspect into AI and ML. There are many deep learning models like Variational Auto Encoders (VAE), Stacked Hourglass networks and Generative Adversarial Networks (GANs) for solving human-image generation. However, it is still difficult for these models to generalize well to the real-world person-image generation task qualitatively. In this paper, we develop a multi-stage model based on Conditional GANs, which could synthesize the new image of the target person given the image of the person and the target (novel) pose desired in real-world scenarios. The model uses 2D keypoints to represent human poses. We propose a Multi-stage Person Generation (MPG) model, in which we have modified the Generator architecture of Pose Guided Person Image Generation (PG2) resulting in two approaches. The first three-stage person generation approach has an additional generator integrated to base architecture and has trained the model end-to-end. The second two-stage person generation approach has a novel texture feature block in stage-II and has been trained incrementally to improve the generation of human-images qualitatively. The proposed two-stage MPG approach has generated promising results on Market-1501 Dataset. The claims are supported by benchmarking the proposed models with recent state-of-the-art models. We also show how the multi-stage conditional GAN architectures influence the process of human-image synthesis. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

DIFFER: Disentangling Identity Features via Semantic Cues for Clothes-Changing Person Re-ID

arXiv

引用

arXiv 2025年

作者： Liang, Xin Rawat, Yogesh S. Center for Research in Computer Vision University of Central Florida United States

Clothes-changing person re-identification (CC-ReID) aims to recognize individuals under different clothing scenarios. Current CC-ReID approaches either concentrate on modeling body shape using additional modalities including silhouette, pose, and body mesh, potentially causing the model to overlook other critical biometric traits such as gender, age, and style, or they incorporate supervision through additional labels that the model tries to disregard or emphasize, such as clothing or personal attributes. However, these annotations are discrete in nature and do not capture comprehensive descriptions. In this work, we propose DIFFER: Disentangle Identity Features From Entangled Representations, a novel adversarial learning method that leverages textual descriptions to disentangle identity features. Recognizing that image features inherently mix inseparable information, DIFFER introduces NBDetach, a mechanism designed for feature disentanglement by leveraging the separable nature of text descriptions as supervision. It partitions the feature space into distinct subspaces and, through gradient reversal layers, effectively separates identity-related features from non-biometric features. We evaluate DIFFER on 4 different benchmark datasets (LTCC, PRCC, CelebreID-Light, and CCVID) to demonstrate its effectiveness and provide state-of-the-art performance across all the benchmarks. DIFFER consistently outperforms the baseline method, with improvements in top-1 accuracy of 3.6% on LTCC, 3.4% on PRCC, 2.5% on CelebReID-Light, and 1% on CCVID. Our code can be found here. © 2025, CC BY.

关键词： Semantics

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：