Deep Learning各种资料网址

徐静

今天轻松写一点，給大家推荐一些Deep Learning的科普的学习资源（作为入门学习），希望对各位有帮助。

1. 人工智能大数据与深度学习 公众号： weic2c

2. 深度学习通俗易懂教程专栏 超智能体 - 知乎专栏：https://zhuanlan.zhihu.com/YJango

3. zouxy9的博客：Deep Learning（深度学习)学习笔记整理，一共八篇，是很基础的内容 http://blog.csdn.net/zouxy09/article/details/8775360/

4. 有趣的机器学习：最简明入门指南 http://blog.jobbole.com/67616/

5. 深度学习如何入门？ http://www.zhihu.com/question/26006703

6. Residual Networks (2015 ICCV, ImageNet 图像分类Top1) 介绍一下2015 ImageNet中分类任务的冠军——MSRA何凯明团队的Residual Networks: http://blog.csdn.net/abcjennifer/article/details/50514124

7. Image Classification with Deep Learning常用模型 Image Vlassification常用的CNN模型，针对cifar-10（for 物体识别），mnist（for 字符识别）& ImageNet（for 物体识别）做一个model 总结 介绍了一下这些网络的结构: http://blog.csdn.net/abcjennifer/article/details/42493493

8. HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection 论文讲解，Faster-rcnn中的proposal提取网络RPN由于特征图的粗糙，在小目标及大IOU阈值情况下的检测率低。论文提出了HyperNet，综合低层，中间层和高层特征获得了较高的recall率: http://blog.csdn.net/cv_family_z/article/details/51135025

9. 有趣的机器学习：最简明入门指南: http://blog.jobbole.com/67616/

10. 目标检测“A MultiPath Network for Object Detection” 对Fast-RCNN方法做了三个小的修改：（1）检测器能够访问多层特征，（2）foveal结构多尺度提取目标上下文信息，（3）在多个IOU下优化损失函数

    http://blog.csdn.net/cv_family_z/article/details/51159619

11. 跟踪“Visual Tracking with Fully Convolutional Networks” 对VGG16特征分析: http://blog.csdn.net/cv_family_z/article/details/50748236

12. Going deeper with convolutions – Googlenet,22层的深度网络。充分利用了网络中的计算资源，通过增加网络的宽度及深度实现。 http://blog.csdn.net/cv_family_z/article/details/50603406

13. SSD: Single Shot MultiBox Detector 本文算是 Faster R-CNN, YOLO 算法的改进版吧，它将检测和分类融合到一起去了，对每个可能的检测框赋予一个类别的概率: http://blog.csdn.net/cv_family_z/article/details/50474679

14. Striving for Simplicity: The All Convolutional Net 全卷积网络: http://blog.csdn.net/cv_family_z/article/details/50403365

15. From Facial Parts Responses to Face Detection: A Deep Learning Approach 公开代码，用CNN进行人脸局部属性检测，然后各个部件综合起来得到人脸检测结果: http://blog.csdn.net/cv_family_z/article/details/50233481

16. 论文提要 Deep Face Recognition,公开代码: http://blog.csdn.net/cv_family_z/article/details/49868979

17. DeepID-Net:multi-stage and deformable deep CNNs for object detection Rcnn改进: http://blog.csdn.net/cv_family_z/article/details/49588969

18. 行人检测“Pedestrian Detection with Unsupervised Multi-Stage Feature Learning”: http://blog.csdn.net/cv_family_z/article/details/49276833

19. 车型识别“Vehicle Type Classification Using a Semisupervised Convolutional Neural Network”: http://blog.csdn.net/cv_family_z/article/details/49154585

20. 论文提要“Learning Deepface Representation”: http://blog.csdn.net/cv_family_z/article/details/48975027

21. 论文提要“Taking a Deeper Look at Pedestrians”:
    http://blog.csdn.net/cv_family_z/article/details/48053535

22. 论文提要“Pedestrian Detection aided by Deep Learning Semantic Tasks”: http://blog.csdn.net/cv_family_z/article/details/47259677

23. 如何简单形象又有趣地讲解神经网络是什么？: http://daily.zhihu.com/story/4424412

24. 深度学习笔记1(卷积神经网络): http://blog.csdn.net/lu597203933/article/details/46575779

25. DeepLearnToolBox中CNN源码解析: http://blog.csdn.net/lu597203933/article/details/46576017

26. CNN(卷积神经网络)、RNN(循环神经网络)、DNN(深度神经网络)的内部网络结构有什么区别: https://www.zhihu.com/question/34681168

27. 针对Faster RCNN具体细节以及源码的解读之SmoothL1Loss层: http://blog.csdn.net/xyy19920105/article/details/50421225

28. 归一化化定义: http://www.cnblogs.com/njustyxy/archive/2011/06/10/2077926.html

29. UFLDL中文教程: http://ufldl.stanford.edu/wiki/index.php/UFLDL教程

30. 介绍:使用卷积神经网络的图像缩放: http://engineering.flipboard.com/2015/05/scaling-convnets/

31. 基于Theano的深度学习(Deep Learning)框架Keras学习随笔-12-核心层: http://blog.csdn.net/niuwei22007/article/details/49277595

32. 如何在Caffe中配置每一个层的结构: http://demo.netfoucs.com/danieljianfeng/article/details/42929283