• 等级
  • 822793 访问
  • 104 原创
  • 1 转发
  • 6659 排名
  • 360 评论
  • 337 获赞

Ubuntu16.04挂载exfat格式U盘

sudoapt-getinstallexfat-fuse

2019-04-02 13:21:41

论文笔记18 --(ReID)Spatial-Temporal Person Re-identification

《Spatial-TemporalPersonRe-identification》论文:https://arxiv.org/abs/1812.03282?context=csGitHub:https://github.com/Wanggcong/Spatial-Temporal-Re-identification这篇是中山大学赖剑煌老师团队发表在AAAI2018上的一篇paper。这篇...

2019-03-19 19:04:19

论文笔记17 --(ReID)SCPNet: Spatial-Channel Parallelism Network for Joint Holistic and Partial Person Re-

《SCPNet:Spatial-ChannelParallelismNetworkforJointHolisticandPartialPersonRe-Identification》论文:https://arxiv.org/pdf/1810.06996.pdfGitHub:https://github.com/xfanplus/Open-SCPNet这是发表在ACCV2...

2019-03-19 18:00:22

论文笔记16 --(ReID)Batch Feature Erasing for Person Re-identification and Beyond

《BatchFeatureErasingforPersonRe-identificationandBeyond》论文:https://arxiv.org/abs/1811.07130GitHub:https://github.com/daizuozhuo/batch-feature-erasing-network类似mgn的切分,而且这篇是站在mgn之上看切分问题的。文章号称...

2019-03-19 17:54:10

论文笔记15 --(ReID)SphereReID: Deep Hypersphere Manifold Embedding for Person Re-Identification

《SphereReID:DeepHypersphereManifoldEmbeddingforPersonRe-Identification》论文:https://arxiv.org/pdf/1704.08063.pdfGitHub:https://github.com/wy1iu/sphereface就是用了个新损失,softmax的变种,但有人指出跟cocoloss是一...

2019-03-19 17:34:01

论文笔记14 --(ReID)Spectral Feature Transformation for Person Re-identification

《SpectralFeatureTransformationforPersonRe-identification》论文:https://arxiv.org/abs/1811.11405这是图森2018年11月提交到arxiv上的一篇文章。用ResNet提取2048-dfeature,然后再对batch内所有image根据视觉相似性(visualsimilarity)建图(这里...

2019-03-17 19:16:20

论文笔记13 -- Performance guarantees for hierarchical clustering

《Performanceguaranteesforhierarchicalclustering》论文:http://cseweb.ucsd.edu/~dasgupta/papers/hier-jcss.pdfGitHub:https://github.com/jonfink/hclusterAbstract作者表示,对于任何度量空间中的任何数据集,都可以构建一个层次聚类,保证对于...

2019-03-13 17:29:41

论文笔记12 -- Deep Residual Learning for Image Recognition

论文:《DeepResidualLearningforImageRecognition》https://arxiv.org/abs/1512.03385https://github.com/KaimingHe/deep-residual-networksAbstract更深的神经网络往往更难以训练,我们在此提出一个残差学习的框架,以减轻网络的训练负担,这是个比以往的网络要深的多...

2019-01-24 17:54:10

论文笔记11 --(ReID)Deep Group-shuffling Random Walk for Person Re-identification

论文:《DeepGroup-shufflingRandomWalkforPersonRe-identification》https://arxiv.org/abs/1807.111784.Experiments4.1.Datasetsandmetric4.2.Implementationdetails我们网络中的成对亲和力CNN采用ResNet-50[11]网...

2019-01-10 11:04:41

论文笔记10 --(ReID)Human Semantic Parsing for Person Re-identification

论文:《HumanSemanticParsingforPersonRe-identification》https://arxiv.org/pdf/1804.00216.pdfAbstract

2019-01-09 17:38:13

PyTorch源码解析--torchvision.transforms(数据预处理、数据增强)

PyTorch框架中有一个很常用的包:torchvisiontorchvision主要由3个子包构成:torchvision.datasets、torchvision.models、torchvision.transforms详细内容可参考:http://pytorch.org/docs/master/torchvision/index.htmlGitHub:https://github.c...

2019-01-04 14:52:48

论文笔记9 -- Residual Attention Network for Image Classification

论文:《ResidualAttentionNetworkforImageClassification》https://arxiv.org/abs/1704.06904https://github.com/fwang91/residual-attention-network这是CVPR2017的一篇论文,在图像分类问题上,首次成功将极深卷积神经网络与人类视觉注意力机制进行有效的结合,...

2019-01-03 18:00:08

Python--numpy和list,查询指定元素的个数及其位置

numpy查询指定元素个数:调用numpy的sum函数>>>importnumpyasnp>>>a=np.array([[0,1,2],[3,4,1]])>>>printaarray([[0,1,2],[3,4,1]])>>>printnp.sum(a==...

2019-01-03 10:47:33

深度学习论文阅读清单

1、(已读)(CVPR2018)《LearningDiscriminativeFeatureswithMultipleGranularitiesforPersonRe-Identification》论文:https://arxiv.org/pdf/1804.01438.pdf笔记:https://blog.csdn.net/lwplwf/article/details/8456...

2018-12-29 16:45:25

论文笔记8 --(ReID)Camera Style Adaptation for Person Re-identification

《CameraStyleAdaptationforPersonRe-identification》论文:https://arxiv.org/abs/1711.10295v1Abstract作为一项跨相机检索任务,由于摄像机的不同,personre-id会收到图像风格变化的影响。在以往的方法中,网络去潜在的学习不受相机风格影响的特征,而本文提出一个camerastyle(Ca...

2018-12-28 14:20:09

论文笔记7 --(ReID)Video-based Person Re-identification via Self Paced Weighting

《Video-basedPersonRe-identificationviaSelfPacedWeighting》,http://mmap.whu.edu.cn/wp-content/uploads/2017/11/aaai-18_wenjun_huang.pdf这是AAAI2018的一篇关于视频PersonRe-ID的论文,作者从自身的步态考虑,将行人的视频序列分成多个子序列,然...

2018-12-28 14:04:01

论文笔记6 --(ReID)A Pose-Sensitive Embedding for Person Re-Id with Expanded Cross Neighborhood Re-Rank

《APose-SensitiveEmbeddingforPersonRe-IdentificationwithExpandedCrossNeighborhoodRe-Ranking》论文:https://arxiv.org/pdf/1711.10378.pdf这是CVPR2018的一篇论文,主要有两个贡献:将姿态信息融入到Re-ID中;设计了一个新的无监督的re-r...

2018-12-25 10:59:15

论文笔记5 --(ReID)Diversity Regularized Spatiotemporal Attention for Video-based Person Re-id

https://arxiv.org/pdf/1803.09882.pdfAbstract

2018-12-20 11:17:04

Ubuntu--切换默认Python版本

可以利用alternatives机制更改Python3为默认:终端执行命令:sudoupdate-alternatives--install/usr/bin/pythonpython/usr/bin/python2100sudoupdate-alternatives--install/usr/bin/pythonpython/usr/bin/python3150...

2018-12-19 15:42:08

论文笔记4 --(ReID)Re-ranking Person Re-identification with k-reciprocal Encoding

《Re-rankingPersonRe-identificationwithk-reciprocalEncoding》论文:https://arxiv.org/abs/1701.08398v1GitHub:https://github.com/zhunzhong07/person-re-ranking这是CVPR2017的一篇paperAbstract当将personre-...

2018-12-06 17:39:27

博客专家

零尾

https://github.com/lwplw
关注
  • 计算机软件/算法工程师
  • 中国 陕西省 西安市
奖章
  • 领英
  • Github
  • 专栏达人