- 2024-10-19(环境篇日志-CVPR2024 ) Physical 3D Adversarial Attacks against Monocular Depth Estimation in Autonomous
题目:Physical3DAdversarialAttacksagainstMonocularDepthEstimationinAutonomousDriving作者:JunhaoZheng,ChenhaoLin*,JiahaoSun,ZhengyuZhao,QianLi,ChaoShen*单位:Xi’anJiaotongUniversity收录:CVPR2024论文:[Physical3DAdversarialAttacks
- 2024-09-06SciTech-Mathmatics-Probability+Statistics: Statistical Inference统计推断- Estimation估计 + Testing Hypothe
轻松学统计:https://zh-cn.statisticseasily.com/词汇表/什么是统计推断/StatisticalInference:SI(统计推断)的类型SI(统计推断)主要有两种类型:Estimation:根据样本数据确定总体的特征;PointEstimation:提供总体参数的单一值估计;ConfidenceInterval:提供
- 2024-08-20[Paper Reading] HandDiff: 3D Hand Pose Estimation with Diffusion on Image-Point Cloud
HandDiff:3DHandPoseEstimationwithDiffusiononImage-PointCloudHandDiff:3DHandPoseEstimationwithDiffusiononImage-PointCloud时间:CVPR2024机构:ETH&CMU&SungkyunkwanUniversityTL;DR将手部分3D点预测任务建模为点云生成任务,提出Handiff算法,以手部
- 2024-08-16[Paper Reading] Single-to-Dual-View Adaptation for Egocentric 3D Hand Pose Estimation
名称Single-to-Dual-ViewAdaptationforEgocentric3DHandPoseEstimation时间:CVPR2024机构:TheUniversityofTokyoTL;DR多目3Dhandposeestimation数据比较难标注,作者核心思路是先训练单目模型,利用无监督的方法适配到双目场景,好处是a.无需标多目数据;b.可以适应任何
- 2024-08-13[Paper Reading] Multiple View Geometry Transformers for 3D Human Pose Estimation
MultipleViewGeometryTransformersfor3DHumanPoseEstimationlink时间:CVPR2024机构:UniversityofToronto&&SoutheastUniversity&&MicrosoftResearchAsiaTL;DR提出一种基于Transformer端到端3DHumanPoseEstimation方法MVGFormer,核心模块是geometry与appea
- 2024-06-05ESSEN: Improving Evolution State Estimation for Temporal Networks using Von Neumann Entropy
我们采用以下六个分类标准:研究重点:这个标准突出了研究的核心目标。网络表示学习旨在找到有效的方法,将复杂的网络结构表示在低维空间中,使其更易于分析并在机器学习任务中使用。例如,Kipf和Welling[7]引入了图卷积网络(GCN)用于静态图上的半监督分类,而Nguyen等人[1
- 2024-06-04Efficient Pruning of Large Language Model with Adaptive Estimation Fusion
本文是LLM系列文章,针对《EfficientPruningofLargeLanguageModelwithAdaptiveEstimationFusion》的翻译。基于自适应估计融合的大型语言模型的高效修剪摘要1引言2相关工作3方法4实验5结论摘要大型语言模型(LLM)已经成为许多生成下游任务的关键,这
- 2023-11-19第二十八周
学习时间:11.13-11.19一、完成内容学习方面:1.看论文《Global-to-LocalModelingforVideo-based3DHumanPoseandShapeEstimation》以及代码实现,做ppt。2.以及关于形变图卷积的论文《DeformableGraphConvolutionalNetworks》和它的应用《Deformablegraphconvolutiona
- 2023-11-16 g2o(5-1)2d-2d
https://github.com/gaoxiang12/slambook/blob/master/ch7/pose_estimation_2d2d.cppCMakeLists.txtcmake_minimum_required(VERSION2.8)project(vo1)set(CMAKE_BUILD_TYPE"Release")set(CMAKE_CXX_FLAGS"-std=c++11-O3")#添加cmake
- 2023-08-29Proj CDeepFuzz Paper Reading: Aries: Efficient Testing of Deep Neural Networks via Labeling-Free Acc
Abstract背景:thedefactostandardtoassessthequalityofDNNsintheindustryistochecktheirperformance(accuracy)onacollectedsetoflabeledtestdatatestselectioncansavelaborandthenbeusedtoassessthemodel前提:themodelshouldhav
- 2023-08-15(十七)MIT公开课雷达系统工程之参数估计与跟踪 (上)
0写在前面只有走在路上,才能摆脱局限、摆脱执着,让所有的选择、探寻、猜测、想象都生机勃勃。——余秋雨《文化苦旅》1基本介绍跟踪雷达:雷达参数估计:准确度、精度和分辨率:2ObservableEstimationObservableAccuracy:LimitationsonRangeEstimation:Theoreticalvs.PracticalAc
- 2023-05-25Density estimation using Real NVP
目录概MotivationRealNVPCouplinglayersDinhL,Sohl-DicksteinJ.andBengioS.Densityestimationusingrealnvp.ICLR,2017.概一种可逆的flow,感觉很diffusion已经非常非常像了.果然,伟大的成果从来不是一蹴而就的.Motivation\(x\)是原始数据,\(z\)
- 2023-05-06Controllable Guarantees for Fair Outcomes via Contrastive Information Estimation
目录概符合说明Motivation优化目标代码GuptaU.,FerberA.M.,DilkinaB.andSteegG.V.Controllableguaranteesforfairoutcomesviacontrastiveinformationestimation.AAAI,2021.概本文提出了一种类似InformationBottleneck的方式用于保证两个群体的fairn
- 2023-03-06EME04 Software effort estimation
SoftwareeffortestimationintroductionDelayestimationuntillateintheprojectBaseestimatesonsimilarprojectsthathavealreadybeencompletedUs
- 2023-03-03Rethinking the Heatmap Regression for Bottom-up Human Pose Estimation
本文的主要思想就是对heatmap图进行一个权重缩放。weight是作者提出的一个思路让模型将低输出值的位置加大权重,高输出值给予小权重,低输出值给与大全中。scaled_gt就是scale
- 2023-02-07RA-Depth: Resolution Adaptive Self-Supervised Monocular Depth Estimation
注:刚入门depthestimation,这也是以后的主要研究方向,欢迎同一个方向的加入QQ群(602708168)交流。1.论文简介论文题目:RA-Depth:ResolutionAdaptiveSelf-SupervisedM
- 2023-01-192022,Feature Evaluation for Underwater Acoustic Object Counting and F0 Estimation
paperAbstract在执行水声目标检测任务时,需要对目标数N进行计数,当N大于1时进行声源分离,并从分离出的噪声中提取每个目标的运动参数(如轴频或FO)。尽管深度学习方法在图像
- 2023-01-18Maximum Likelihood Estimation极大似然估计
WelcomeToMyBlog贝叶斯公式现通过分类问题解释贝叶斯公式:P(c)是类别c的先验(prior)概率P(x|c)是似然概率(likelihoodprobability),或者说是样本x相对于类