BYR Achieve · 镜像论坛

# 前言本文主要对该博客的文章进行汇总和分类，因为博客规模逐渐变大之后，索引文章就变成一个日渐凸显的问题了，变得难以维护起来。本博客主要的原创文章将在公众号与知乎专栏同步更新，有兴趣的朋友可以同步关注。 $\nabla$ 联系方式： **e-mail**: FesianXu@gmail.com **QQ**: 973926198 **github**: https://github.com/FesianXu **知乎专栏**: [计算机视觉/计算机图形理论与应用](https://zhuanlan.zhihu.com/c_1265262560611299328) 微信公众号：https://img-blog.csdnimg.cn/20201118234623868.png#pic_center ------ # 图神经网络相关 1. [《学习geometric deep learning笔记系列》第一篇，Non-Euclidean Structure Data之我见](https://blog.csdn.net/LoseInVain/article/details/88373506) 2. [《Geometric Deep Learning学习笔记》第二篇，在Graph上定义卷积操作，图卷积网络](https://blog.csdn.net/LoseInVain/article/details/90171863) 3. [《Geometric Deep Learning学习笔记》第三篇，GCN的空间域理解，Message Passing以及其含义](https://blog.csdn.net/LoseInVain/article/details/90348807) 4. [Shift-GCN网络论文笔记](https://fesian.blog.csdn.net/article/details/109563113) 5. [Shift-GCN中Shift的实现细节笔记，通过torch.index_select实现](https://fesian.blog.csdn.net/article/details/109644297) # 卷积神经网络优化与加速，压缩 1. [紧致卷积网络设计——Shift卷积算子](https://fesian.blog.csdn.net/article/details/109474701) 2. [卷积网络模型压缩的若干总结](https://blog.csdn.net/LoseInVain/article/details/96651171) 3. [《weekly paper》DenseNet的理解](https://blog.csdn.net/LoseInVain/article/details/80453996) 4. [一文搞懂反卷积，转置卷积](https://fesian.blog.csdn.net/article/details/81098502) # 人体活动与视频分析 1. [基于图结构的视频理解——组织视频序列的非线性流](https://fesian.blog.csdn.net/article/details/108212429) 2. [万字长文漫谈视频理解](https://fesian.blog.csdn.net/article/details/105545703) 3. [【动作识别相关，第一篇】skeleton骨骼点数据类型介绍](https://fesian.blog.csdn.net/article/details/87901764) 4. [[笔记] 常见人体铰链关节点数据集中的关节点排序(SMPL,NTU,MPII,human3.6M)](https://fesian.blog.csdn.net/article/details/108242717) # 立体视觉与多视角视觉 1. [双目三维重建——层次化重建思考](https://fesian.blog.csdn.net/article/details/107720442) 2. [【多视角立体视觉系列】几何变换的层次——投影变换，仿射变换，度量变换和欧几里德变换](https://fesian.blog.csdn.net/article/details/104533575) 3. [【多视角立体视觉系列】 conic圆锥线和quadric二次曲锥面的定义和应用](https://fesian.blog.csdn.net/article/details/104515839) 4. [讨论物体的表面深度对相机成像的影响](https://fesian.blog.csdn.net/article/details/102869987) 5. [图像校正(Image Rectification)——使得在对极线上寻找对应点更加容易](https://fesian.blog.csdn.net/article/details/102775734) 6. [几何变换——关于透视变换和仿射变换以及齐次坐标系的讨论](https://fesian.blog.csdn.net/article/details/102756630) 7. [立体视觉中的对极几何——如何更好更快地寻找对应点](https://fesian.blog.csdn.net/article/details/102665911) 8. [论相机中心投影中，相机中心的作用](https://fesian.blog.csdn.net/article/details/103369203) 9. [投影相机，透视相机，弱透视相机和仿射相机的区别和联系](https://fesian.blog.csdn.net/article/details/102883243) 10. [相机中的透视投影几何——讨论相机中的正交投影，弱透视投影以及透视的一些性质](https://fesian.blog.csdn.net/article/details/102698703) 11. [相机的针孔模型及其内参数，外参数的理解](https://fesian.blog.csdn.net/article/details/102632940) 12. [基于匹配点集对单应性矩阵进行估计](https://fesian.blog.csdn.net/article/details/105915325) 13. [从手写字符匹配开始，简要解释局部仿射变换(local affine transformation)](https://fesian.blog.csdn.net/article/details/108454304) # 深度学习框架相关 `tensorflow` 1. [tf.gather, tf.gather_nd和tf.slice](https://fesian.blog.csdn.net/article/details/85339985) 2. [tensorflow编程实践：结构化你的模型](https://fesian.blog.csdn.net/article/details/82085185) 3. [TensorFlow的体系结构](https://fesian.blog.csdn.net/article/details/81952515) 4. [tensorflow中的image预处理操作函数](https://fesian.blog.csdn.net/article/details/81774840) 5. [tensorflow中的位操作](https://fesian.blog.csdn.net/article/details/81750158) 6. [tf.tuple()用于组合多个张量输入](https://fesian.blog.csdn.net/article/details/81704072) 7. [tf.group()用于组合多个操作](https://fesian.blog.csdn.net/article/details/81703786) 8. [TensorFlow高阶函数之 tf.foldl()和tf.foldr()](https://fesian.blog.csdn.net/article/details/81635711) 9. [tf.nn.softmax_cross_entropy_with_logits 将在未来弃用](https://fesian.blog.csdn.net/article/details/80932605) 10. [TensorFlow和Keras中的Crop函数](https://fesian.blog.csdn.net/article/details/80488645) 11. [如何在TensorFlow中使用并行数据加载，解决视频读取问题](https://fesian.blog.csdn.net/article/details/80311534) 12. [PLSTM的TensorFlow实现与解释](https://fesian.blog.csdn.net/article/details/79862072) 13. [TensorFlow中的LSTM源码理解与二次开发](https://fesian.blog.csdn.net/article/details/79642721) 14. [tf.concat, tf.stack和tf.unstack的用法](https://fesian.blog.csdn.net/article/details/79638183) 15. [tf.squeeze()用于压缩张量中为1的轴](https://fesian.blog.csdn.net/article/details/78994695) 16. [tf.transpose()交换张量位置（矩阵转置）](https://fesian.blog.csdn.net/article/details/78994660) 17. [tf.nn.conv2d()使用](https://fesian.blog.csdn.net/article/details/78935192) 18. [利用numpy数组保存TensorFlow模型的参数](https://fesian.blog.csdn.net/article/details/78935157) 19. [Effective TensorFlow Chapter 9: TensorFlow模型原型的设计和利用python ops的高级可视化](https://fesian.blog.csdn.net/article/details/78867452) 20. [Effective TensorFlow Chapter 8: 在TensorFlow中的控制流：条件语句和循环](https://fesian.blog.csdn.net/article/details/78867004) 21. [Effective TensorFlow Chapter 7: TensorFlow中的执行顺序和控制依赖](https://fesian.blog.csdn.net/article/details/78780020) 22. [Effective TensorFlow Chapter 6: 在TensorFlow中的运算符重载](https://fesian.blog.csdn.net/article/details/78779182) 23. [Effective TensorFlow Chapter 5: 在TensorFlow中，给模型喂数据(feed data)](https://fesian.blog.csdn.net/article/details/78769879) 24. [Effective TensorFlow Chapter 4: TensorFlow中的广播Broadcast机制](https://fesian.blog.csdn.net/article/details/78763303) 25. [Effective TensorFlow Chapter 3: 理解变量域Scope和何时应该使用它](https://fesian.blog.csdn.net/article/details/78762816) 26. [Effective TensorFlow Chapter 2: 理解静态和动态的Tensor类型的形状](https://fesian.blog.csdn.net/article/details/78762739) 27. [tf.one_hot()进行独热编码](https://fesian.blog.csdn.net/article/details/78819390) 28. [TensorFlow中的高阶函数：tf.map_fn()](https://fesian.blog.csdn.net/article/details/78815130) 29. [tf.device()指定运行设备](https://fesian.blog.csdn.net/article/details/78814091) 30. [TensorFlow模型的保存和持久化](https://fesian.blog.csdn.net/article/details/78241000) 31. [TensorlFlow中的一些坑](https://fesian.blog.csdn.net/article/details/78150427) 32. # 深度学习框架相关 `pytorch` 1. [Shift-GCN中Shift的实现细节笔记，通过torch.index_select实现](https://fesian.blog.csdn.net/article/details/109644297) 2. [测量pytorch代码段的运行时间](https://fesian.blog.csdn.net/article/details/106713550) 3. [在pytorch中停止梯度流的若干办法，避免不必要模块的参数更新](https://fesian.blog.csdn.net/article/details/105461904) 4. [在pytorch中对非叶节点的变量计算梯度](https://fesian.blog.csdn.net/article/details/99172594) 5. [pytorch手动实现滑动窗口操作，论fold和unfold函数的使用](https://fesian.blog.csdn.net/article/details/88139435) 6. [在pytorch中动态调整优化器的学习率](https://fesian.blog.csdn.net/article/details/87858408) 7. [Pytorch的BatchNorm层使用中容易出现的问题](https://fesian.blog.csdn.net/article/details/86476010) 8. [[临时笔记] pytorch报错消息及其解决纪录](https://fesian.blog.csdn.net/article/details/86140412) 9. [在pytorch中进行预训练模型的加载和模型的fine-tune操作](https://fesian.blog.csdn.net/article/details/84982332) 10. [在TensorFlow中自定义梯度的两种方法](https://fesian.blog.csdn.net/article/details/83108001) 11. [《临时笔记》用pytorch踩过的坑](https://fesian.blog.csdn.net/article/details/82916163) 12. [pytorch中的L2和L1正则化，自定义优化器设置等操作](https://fesian.blog.csdn.net/article/details/81708474) 13. [在pytorch中的双线性采样（Bilinear Sample）](https://fesian.blog.csdn.net/article/details/108732249) 14. [<深度学习系列>基于numpy和python的反向传播算法的实现与分析](https://fesian.blog.csdn.net/article/details/78239068) 15. # 深度学习框架 `caffe2`和`caffe` 1. [《临时笔记》caffe2使用记录](https://fesian.blog.csdn.net/article/details/105863809) # C语言与底层原理 1. [C语言中去除不必要的内存引用可以有效地提高性能](https://fesian.blog.csdn.net/article/details/104150155) 2. [c语言中内循环和外循环的位置可能产生性能上的区别](https://fesian.blog.csdn.net/article/details/104129412) 3. [[C语言朝花夕拾] C语言中的命令行输入参数判断](https://fesian.blog.csdn.net/article/details/109744807) 4. [用“位操作”取代“取模操作”判断奇数偶数](https://fesian.blog.csdn.net/article/details/105228116) 5. [c语言运行时出现segment fault的原因](https://fesian.blog.csdn.net/article/details/104127334) 6. [一文理解C语言中的volatile修饰符](https://fesian.blog.csdn.net/article/details/103356324) 7. [C语言中的内存布局(memory layout)](https://fesian.blog.csdn.net/article/details/103183829) 8. # CUDA编程相关 1. [[临时笔记] cuda报错日常](https://fesian.blog.csdn.net/article/details/109488832) # 深度学习系统搭建 1. [[darknet源码系列-1] darknet源码中的常见数据结构](https://fesian.blog.csdn.net/article/details/109779812) 2. [数据并行和模型并行的区别](https://fesian.blog.csdn.net/article/details/105808818) 3. [Conv卷积层的反向求导细节](https://fesian.blog.csdn.net/article/details/103363185) 4. [基于代码的Pooling池化层的反向求导细节](https://fesian.blog.csdn.net/article/details/98451913) 5. [《AutoDiff理解》之第一篇，自动求导技术在深度学习中的应用](https://fesian.blog.csdn.net/article/details/88557173) 6. # 机器学习原理 1. [数据，模型，算法共同决定深度学习模型效果](https://fesian.blog.csdn.net/article/details/105644994) 2. [一文理解Ranking Loss/Contrastive Loss/Margin Loss/Triplet Loss/Hinge Loss](https://fesian.blog.csdn.net/article/details/103995962) 3. [参数和非参数模型——当我谈到参数我在说些什么](https://fesian.blog.csdn.net/article/details/103951311) 4. [在深度学习中，对于特征融合方式的思考——论pointwise addition和concatenate的异同](https://fesian.blog.csdn.net/article/details/88363776) 5. [损失函数的可视化——浅论模型的参数空间与正则](https://fesian.blog.csdn.net/article/details/83473975) 6. [曲线拟合问题与L2正则](https://fesian.blog.csdn.net/article/details/82824987) 7. [贝叶斯曲线拟合以及对L2正则化的贝叶斯解释](https://fesian.blog.csdn.net/article/details/82822631) 8. [生成模型和判别模型的区别](https://fesian.blog.csdn.net/article/details/82785985) 9. [分类问题的两大过程，推理和决策](https://fesian.blog.csdn.net/article/details/82789501) 10. [贝叶斯决策](https://fesian.blog.csdn.net/article/details/82780472) 11. [概率学派和贝叶斯学派的区别](https://fesian.blog.csdn.net/article/details/80499147) 12. [理解多维高斯分布](https://fesian.blog.csdn.net/article/details/80339201) 13. [Logistic regression(逻辑斯蒂回归)](https://fesian.blog.csdn.net/article/details/79411244) 14. [在机器学习中epoch, iteration, batch_size的区别](https://fesian.blog.csdn.net/article/details/79348965) 15. [经验误差，泛化误差](https://fesian.blog.csdn.net/article/details/78746520) 16. [<深度学习系列>深度学习中激活函数的选择](https://fesian.blog.csdn.net/article/details/78109098) 17. [机器学习性能评估指标](https://fesian.blog.csdn.net/article/details/78109029) 18. [机器学习模型的容量，过拟合与欠拟合](https://fesian.blog.csdn.net/article/details/78108990) 19. [训练集，测试集，检验集的区别与交叉检验](https://fesian.blog.csdn.net/article/details/78108955) 20. [《深度学习系列》反向传播算法的公式推导](https://fesian.blog.csdn.net/article/details/78092613) 21. [机器学习系列之感知器模型](https://fesian.blog.csdn.net/article/details/78430585) 22. [＜机器学习系列＞线性回归模型](https://fesian.blog.csdn.net/article/details/78245665) 23. [随机梯度下降法，批量梯度下降法和小批量梯度下降法以及代码实现](https://fesian.blog.csdn.net/article/details/78243051) 24. [梯度下降法求函数最小值基于matlab实现](https://fesian.blog.csdn.net/article/details/52804517) # 人体动捕相关 1. [视频人体动作捕捉技术](https://fesian.blog.csdn.net/article/details/108322781) 2. [人体动作捕捉与SMPL模型 (mocap and SMPL model)](https://fesian.blog.csdn.net/article/details/107265821) 3. # 计算机图形学相关 1. [薄板样条插值(Thin Plate Spline)](https://fesian.blog.csdn.net/article/details/108483736) 2. [从手写字符匹配开始，简要解释局部仿射变换(local affine transformation)](https://fesian.blog.csdn.net/article/details/108454304) 3. [[GAMES101学习笔记] 角度与立体角](https://fesian.blog.csdn.net/article/details/108630648) # 深度学习trick 1. [深度学习debug沉思录第二集](https://fesian.blog.csdn.net/article/details/86560548) 2. [深度学习debug沉思录](https://fesian.blog.csdn.net/article/details/83021356) 3. # SVM系列 1. [《SVM笔记系列之一》什么是支持向量机SVM](https://fesian.blog.csdn.net/article/details/78636176) 2. [《SVM笔记系列之二》SVM的拉格朗日函数表示以及其对偶问题](https://fesian.blog.csdn.net/article/details/78636285) 3. [《SVM笔记系列之三》拉格朗日乘数法和KKT条件的直观解释](https://fesian.blog.csdn.net/article/details/78624888) 4. [《SVM笔记系列之四》最优化问题的对偶问题](https://fesian.blog.csdn.net/article/details/78636341) 5. [《SVM笔记系列之五》软间隔线性支持向量机](https://fesian.blog.csdn.net/article/details/78646479) 6. [《SVM笔记系列之六》支持向量机中的核技巧那些事儿](https://blog.csdn.net/LoseInVain/article/details/83097093) # 图片动画化技术 1. [运动的零阶分解与一阶分解以及在图片动画化中的应用 I](https://fesian.blog.csdn.net/article/details/108710063) 2. # 网络工程 1. [<笔记>MTU和MSS的区别](https://fesian.blog.csdn.net/article/details/53694265) 2. [<理解共享信道访问协议系列1>共享信道前世今生](https://fesian.blog.csdn.net/article/details/52645324) # Linux运维与深度学习环境搭建 1. [[linux命令] 如何在用户终端退出后，不挂起或退出用户正在运行的程序](https://fesian.blog.csdn.net/article/details/102534989) 2. [什么是DevOps](https://fesian.blog.csdn.net/article/details/106814229) 3. [利用远程服务器实现内网穿透访问jupyter notebook](https://fesian.blog.csdn.net/article/details/101875110) 4. [[Linux配置笔记] vimplus的配置过程及其报错纪录](https://fesian.blog.csdn.net/article/details/96211291) 5. [[linux常用命令] rsync 用于远程/本地文件的拷贝（可以实现差量复制）](https://fesian.blog.csdn.net/article/details/88131041) 6. [深度学习常用软件纪录](https://fesian.blog.csdn.net/article/details/86496061) 7. [ubuntu文件系统常用文件备忘录](https://fesian.blog.csdn.net/article/details/86190076) 8. [在linux系统中ftp或者docker的数据卷中使用mount --bind](https://fesian.blog.csdn.net/article/details/85056401) 9. [linux中常用的用户与用户组相关命令](https://fesian.blog.csdn.net/article/details/84978089) 10. [在linux系统上部署FTP服务时进行权限管理(利用chown,chmod命令实现)](https://fesian.blog.csdn.net/article/details/84638932) # Python相关 1. [python修饰器教程](https://fesian.blog.csdn.net/article/details/82055524) 2. [einsum的基础使用](https://fesian.blog.csdn.net/article/details/81143966) 3. # 物体检测算法 1. [Object Detection中的mAP](https://fesian.blog.csdn.net/article/details/106442683) 2. [《土豆学Object Detection》之 RCNN初探](https://fesian.blog.csdn.net/article/details/98054030) 3. # 数据融合与数据处理相关 1. [一文理清卡尔曼滤波，从传感器数据融合开始谈起](https://fesian.blog.csdn.net/article/details/90340087) 2. # 其他 1. [markdown数学公式编辑](https://fesian.blog.csdn.net/article/details/78747397) 2. [设置可见GPU，进行多显卡深度学习训练](https://fesian.blog.csdn.net/article/details/78146459) 3. [什么是DevOps](https://fesian.blog.csdn.net/article/details/106814229) 4. [《临时笔记》编程生涯中的傻傻的bug问题](https://fesian.blog.csdn.net/article/details/105795982) 5. [在vs code上进行远程深度学习开发环境简易搭建](https://fesian.blog.csdn.net/article/details/103872261) 6. [《临时笔记》一些深度学习中的英文术语的纪录](https://fesian.blog.csdn.net/article/details/103870157) 7. [SIFT,SUFT等具有知识产权的算法将不能在opencv4中使用](https://fesian.blog.csdn.net/article/details/102945633) 8. [《临时笔记》一些计算机视觉的英语术语的纪录](https://fesian.blog.csdn.net/article/details/102739778) 9. [《临时笔记》编程生涯中的傻傻的bug问题](https://fesian.blog.csdn.net/article/details/105795982)

【心得】机器学习杂货铺的汇总贴