![](https://img-blog.csdnimg.cn/20201014180756780.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
深度学习
Arron_hou
热爱学习的小透明!
展开
-
【Optimizor】Pytorch 实现汇总
Optimizor定义:fθf_\thetafθ表示模型θt\theta_tθt 表示模型在更新t次后的参数y^=fθ(x)\hat{y} = f_\theta(x)y^=fθ(x) 表示模型预测值J(y^,y∗)J(\hat{y},y^*)J(y^,y∗) 表示真值 y∗y^*y∗ 与预测值 y^\hat{y}y^ 的损失函数∇J(θt)\nabla J\left(\t...原创 2020-05-06 15:46:29 · 389 阅读 · 0 评论 -
【Backbone】IBN-Net
IBN-NetTwo at Once: Enhancing Learning and Generalization Capacities via IBN-Nethttps://arxiv.org/abs/1807.09441Xingang Pan1, Ping Luo1, Jianping Shi2, and Xiaoou Tang1出自商汤, ECCV2018Question如何提高...原创 2020-05-03 15:24:13 · 695 阅读 · 0 评论 -
【ReID】首届“全国人工智能大赛”(行人重识别 Person ReID 赛项)
首届“全国人工智能大赛”(行人重识别 Person ReID 赛项原创 2020-04-15 00:27:05 · 2259 阅读 · 1 评论 -
【Continual Learning】Encoder Based Lifelong Learning
Encoder Based Lifelong LearningICCV2017Amal Rannen Triki ∗† Rahaf Aljundi∗ Mathew B. Blaschko Tinne Tuytelaars KU LeuvenKU Leuven,ESAT-PSI, iMinds, BelgiumQuestionOur method aims at preserving...原创 2020-04-13 18:02:37 · 1011 阅读 · 1 评论 -
【Backbone】resNet
resnetDeep Residual Learning for Image Recognition (2015-12)https://arxiv.org/abs/1512.03385ResNet 最重要的贡献就是short cut,通过增加恒等映射,解决模型退化的问题。为什么可以解决模型退化问题呢?首先什么是模型退化问题,也就是加深简单的模型,不能直接提升模型的性能。我们在训练模...原创 2019-08-16 12:09:06 · 740 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- Encoder And Decoder
【Semantic Segmentation】语义分割综述--Encoder And DecoderEncoder And Decoder[FCN] Fully Convolutional Networks for Semantic Segmentation 2016-05FCN-32FCN-16FCN-8code by pytorch[U-Net] Convolutional Networks...原创 2019-09-20 18:24:57 · 2379 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- Pyramid & Multi-Path
【Semantic Segmentation】语义分割综述 --- Pyramid & Multi-PathPyramid & Multi-Path[RefineNet] Multi-Path Refinement Networks for High-Resolution Semantic Segmentation 2016-11[PSPNet] Pyramid Scene Par...原创 2019-09-20 18:28:37 · 388 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- Atrous Convolution
【Semantic Segmentation】语义分割综述 -- Atrous ConvolutionAtrous Convolution[DeepLabV2] Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution,and Fully Connected CRFs 2016[ASPP] atrous...原创 2019-09-20 18:33:00 · 548 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- Attention
【Semantic Segmentation】语义分割综述 -- AttentionAttention[Attention U-Net] Learning Where to Look for the Pancreas( MIDL 2018)[PAN] Pyramid Attention Network for Semantic Segmentation (BMVC 2018)[PSANet] Po...原创 2019-09-20 18:35:11 · 2645 阅读 · 1 评论 -
【Semantic Segmentation】语义分割综述 -- Faster
【Semantic Segmentation】语义分割综述 -- FasterFaster[ESPNet] Efficient Spatial Pyramid of Dilated Convolutions for Semantic Segmentation 2018-03[Fast-SCNN] Fast Semantic Segmentation Network 2019-02[BiSeNet]...原创 2019-09-20 18:36:40 · 446 阅读 · 1 评论 -
【Semantic Segmentation】语义分割综述 - - Semi-supervised
【Semantic Segmentation】语义分割综述 - - Semi-supervisedSemi-supervisedWeakly-Supervised Semantic Segmentation by Iteratively Mining Common Object Features CVPR18Weakly-Supervised Semantic Segmentation Netwo...原创 2019-09-20 18:39:42 · 686 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- NAS
【Semantic Segmentation】语义分割综述 -- NASNAS[Auto-DeepLab] Hierarchical Neural Architecture Search for Semantic Image Segmentation (CVPR2019)NAS[Auto-DeepLab] Hierarchical Neural Architecture Search for ...原创 2019-09-20 18:41:09 · 569 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述 -- loss
【Semantic Segmentation】语义分割综述 -- losslossohemfocal-loss[Lovasz-Softmax loss] A tractable surrogate for the optimization of theintersection-over-union measure in neural networks (CVPR 2018)lossohemf...原创 2019-09-20 18:47:01 · 573 阅读 · 0 评论 -
【Semantic Segmentation】语义分割综述
【Semantic Segmentation】语言分割综述包括encoder decoder 结构空洞卷积注意力机制等在图像分割中的应用原创 2019-08-22 11:10:41 · 2289 阅读 · 0 评论 -
【Attention】注意力机制在图像上的应用
【Attention】注意力机制在图像上的应用原创 2019-07-12 23:27:24 · 7013 阅读 · 0 评论 -
kaggle-组织病理学癌症检测
组织病理学癌症检测Histopathologic Cancer DetectionProject understanding比赛链接:https://www.kaggle.com/c/histopathologic-cancer-detection原文参考: https://www.kaggle.com/qitvision/a-complete-ml-pipeline-fast-ai识别从较...原创 2019-03-01 14:00:09 · 4649 阅读 · 0 评论 -
李宏毅ML lecture-1,2,3 线性回归及梯度下降
李宏毅ML lecture-2,3 线性回归及梯度下降 线性回归(Linear regression)损失函数( loss function)平方损失(square loss)学习率(Learning rate)momentumdecaynesterovRMSpropAdagradAdadeltaAdamAdamaxNadam偏执bias和方差variance正则化(regularization)...原创 2019-02-21 00:56:47 · 385 阅读 · 0 评论 -
李宏毅ML lecture-10 CNN
李宏毅ML lecture-10 CNNConvolutionstridepaddingMaxPoolingflattenFully Connected FeedforwardCNN的直观理解相关论文理解CNN深度强化网络CVAlexNetR-CNNNLP语言识别情感分析其他论文CNN比DNN更为简单CNN过滤器,目的就是寻找一种pattern,这种pattern远远小于整个图片的大小,同时是...原创 2019-02-21 00:58:55 · 602 阅读 · 0 评论 -
李宏毅ML lecture-11 why deep
李宏毅ML lecture-11 why deepDeep → Modularization电路窗花End-to-end Learning使用Deep Learning 的理由learning more这节课主要将深度学习的深度是如何体现的.Seide, Frank, Gang Li, and Dong Yu. “Conversational Speech Transcription Usin...原创 2019-02-21 00:59:38 · 303 阅读 · 0 评论 -
李宏毅ML lecture-14 unsupervised Learning--Word Embeddng
李宏毅ML lecture-14 unsupervised Learning--Word Embeddng词嵌入(Word Embeddng),让计算机理解文章.实现方法有两种:https://nlp.stanford.edu/projects/glove/word vector paperTomas Mikolovhttps://blog.csdn.net/m0epnwsty...原创 2019-02-21 01:01:44 · 436 阅读 · 0 评论 -
李宏毅ML lecture-17-18 unsupervised Learning Generation
李宏毅ML lecture-16 unsupervised Learning Generation原创 2019-03-09 14:40:48 · 509 阅读 · 0 评论 -
Deep Neural Networks for Object Detection
Deep Neural Networks for Object Detection论文介绍1) 论文来源Szegedy C, Toshev A, Erhan D. Deep neural networks for object detection[C]//Advances in neural information processing systems. 2013: 2553-2561.2...原创 2019-04-17 01:23:11 · 765 阅读 · 0 评论 -
【Object Detection】目标检测综述
目标检测论文介绍主要内容Deep Neural Networks for Object Detection论文介绍主要内容OverFeat,Integrated Recognition, Localization and Detection using Convolutional Networks.Rich feature hierarchies for accurate object detec...原创 2019-09-02 03:01:35 · 876 阅读 · 0 评论 -
OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks
OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks原创 2019-04-18 01:04:48 · 325 阅读 · 0 评论 -
【Pose】Convolutional Pose Machines
【Pose】Convolutional Pose Machines3.1 Method3.2 Convolutional Pose Machines:3.2.1 Keypoint Localizaton Using Local Image Evidence3.2.2 Sequential Prediction with Learned Spatial Context Features算法流程使用阶...原创 2019-06-02 19:01:27 · 563 阅读 · 0 评论 -
李宏毅ML lecture-13 unsupervised Learning Principle Component Analysis
李宏毅ML lecture-13 unsupervised Learning Principle Component Analysis聚类(Clustering)K-meansHAC数据降维(Dimension Reduction)PCA(Principal Component Analysis)基于特征值分解协方差矩阵实现PCA算法特征向量拉格朗日乘数法基于SVD分解协方差矩阵实现PCA算法PC...原创 2019-02-25 12:10:22 · 596 阅读 · 0 评论