【OpenFace】

超、凢脫俗 2022-06-12 13:23 132阅读 0赞

OpenFace：

[http://cmusatyalab.github.io/openface/][http_cmusatyalab.github.io_openface]

# 一、什么是Openface？ #

Openface是一个基于深度神经网络的开源人脸识别系统。该系统基于谷歌的文章[FaceNet: A Unified Embedding for Face Recognition and Clustering][FaceNet_ A Unified Embedding for Face Recognition and Clustering]。Openface是卡内基梅隆大学的[Brandon Amos][]主导的。

官方地址：[http://cmusatyalab.github.io/openface/][http_cmusatyalab.github.io_openface]

代码：[https://github.com/cmusatyalab/openface][https_github.com_cmusatyalab_openface]

# 二、Openface环境搭建 #

系统：Ubuntu 14.04 64位桌面[操作系统][Link 1]

参考：[http://cmusatyalab.github.io/openface/setup/][http_cmusatyalab.github.io_openface_setup]

### 1、Ubuntu切换root用户 ###

此处不详述，如果要用普通用户，请自行[测试][Link 2]。

参考文章：

[http://blog.csdn.net/itas109/article/details/50679251][http_blog.csdn.net_itas109_article_details_50679251]

### 2、安装前准备工作 ###

安装必要的程序，可以用下面的批处理，也可以一个一个的进行安装。

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  \#!/bin/sh
2.  sudo apt-get install build-essential -y 
3.  sudo apt-get install cmake -y 
4.  sudo apt-get install curl -y 
5.  sudo apt-get install gfortran -y 
6.  sudo apt-get install git -y 
7.  sudo apt-get install libatlas-dev -y 
8.  sudo apt-get install libavcodec-dev -y 
9.  sudo apt-get install libavformat-dev -y 
10. sudo apt-get install libboost-all-dev -y 
11. sudo apt-get install libgtk2.0-dev -y 
12. sudo apt-get install libjpeg-dev -y 
13. sudo apt-get install liblapack-dev -y 
14. sudo apt-get install libswscale-dev -y 
15. sudo apt-get install pkg-config -y 
16. sudo apt-get install python-dev -y 
17. sudo apt-get install python-pip -y 
18. sudo apt-get install wget -y 
19. sudo apt-get install zip –y

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

### 3、安装必要的库 ###

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  pip2 install numpy scipy pandas 
2.  pip2 install scikit-learn scikit-image

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  pip2 install numpy scipy pandas 
2.  pip2 install scikit-learn scikit-image

注意：

a.如果出现某一个安装失败的情况，可以一个一个的安装

b.提高pip安装速度

可以更换pip镜像加快下载速度  
建立./pip/pip.conf，输入以下内容（或者其他可用镜像）：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  \[global\] 
2.  timeout = 6000 
3.  index-url = http://pypi.douban.com/simple
4.  
5.  \[install\] 
6.  use-mirrors = **true**
7.  mirrors = <a target=\_blank href="http://pypi.douban.com/">http://pypi.douban.com/</a>

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

c.报错：SSLError: The read operation timed out

可以用下列指令将延时加长

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  pip2 -install scikit-image --timeout 100

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  pip2 -install scikit-image --timeout 100

### 4、安装Torch ###

a.安装依赖

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  curl -shttps://raw.githubusercontent.com/torch/ezinstall/master/install-deps | bash –e

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  curl -shttps://raw.githubusercontent.com/torch/ezinstall/master/install-deps | bash –e

b.安装

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git clone https://github.com/torch/distro.git ~/torch --recursive
2.  cd ~/torch && ./install.sh

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git clone https://github.com/torch/distro.git ~/torch --recursive
2.  cd ~/torch && ./install.sh

c.安装依赖

luarocks install $NAME, where `$NAME` is as listed below.

[**dpnn**][dpnn]

[**nn**][nn]

[**csvigo**][csvigo]

[**cunn**][cunn] **(使用CUDA)**

[**fblualib**][fblualib] **(仅为了训练**[**DNN**][DNN]**)**

[**torchx**][torchx] **(仅为了训练**[**DNN**][DNN]**)**

命令行，按照需要安装：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  ~/torch/install/bin/luarocks install dpnn 
2.  ~/torch/install/bin/luarocks install nn 
3.  ~/torch/install/bin/luarocks install optim 
4.  ~/torch/install/bin/luarocks install csvigo 
5.  ~/torch/install/bin/luarocks install cunn 
6.  ~/torch/install/bin/luarocks install fblualib 
7.  ~/torch/install/bin/luarocks install torchx

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

d.验证是否安装依赖成功

用th命令验证

注意：

a.gitclone更新网络老中断

[Git][] submodule update --init –recursive

或者torch目录下的

Update.sh

建议用Update.sh解决

b.错误：

Cloning into'extra/luaffifb'...

remote:Counting objects: 918, done.

error: RPCfailed; result=56, HTTP code = 200| 0 bytes/s

fatal: Theremote end hung up unexpectedly

fatal: earlyEOF

fatal:index-pack failed

Clone of 'https://github.com/facebook/luaffifb' intosubmodule path 'extra/luaffifb' failed

解决：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git config --global http.postBuffer 524288000

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git config --global http.postBuffer 524288000

### 5、安装opencv ###

[OpenCV][]版本为2.4.11，下载地址：[https://github.com/Itseez/opencv/archive/2.4.11.zip][https_github.com_Itseez_opencv_archive_2.4.11.zip]

编译参考：[http://docs.opencv.org/2.4/doc/tutorials/introduction/linux\_install/linux\_install.html][http_docs.opencv.org_2.4_doc_tutorials_introduction_linux_install_linux_install.html]

a.指令下载：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  cd ~ 
2.  mkdir -p src 
3.  cd src 
4.  curl -L https://github.com/Itseez/opencv/archive/2.4.11.zip -o ocv.zip

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  cd ~ 
2.  mkdir -p src 
3.  cd src 
4.  curl -L https://github.com/Itseez/opencv/archive/2.4.11.zip -o ocv.zip

b.解压：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  unzip ocv.zip

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  unzip ocv.zip

c.编译：

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  cd opencv-2.4.11 
2.  mkdir release 
3.  cd release 
4.  cmake -D CMAKE\_BUILD\_TYPE=RELEASE -D CMAKE\_INSTALL\_PREFIX=/usr/local .. 
5.  make -j8 
6.  make install

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  cd opencv-2.4.11 
2.  mkdir release 
3.  cd release 
4.  cmake -D CMAKE\_BUILD\_TYPE=RELEASE -D CMAKE\_INSTALL\_PREFIX=/usr/local .. 
5.  make -j8 
6.  make install

###  ###

d.验证

import cv2

###    6、安装dlib ###

dlib v18.16下载地址：[https://github.com/davisking/dlib/releases/download/v18.16/dlib-18.16.tar.bz2][https_github.com_davisking_dlib_releases_download_v18.16_dlib-18.16.tar.bz2]

a.安装编译

**\[html\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  mkdir -p ~/src 
2.  cd ~/src tar xf dlib-18.16.tar.bz2 
3.  cd dlib-18.16/python\_examples 
4.  mkdir build 
5.  cd build 
6.  cmake ../../tools/python 
7.  cmake --build . --config Release 
8.  cp dlib.so /usr/local/lib/python2.7/dist-packages

**\[html\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

b.确保

在a中最后一条命令中，确保路径在默认的[Python][]路径，可以在[Python][Python 1]解释器里面用sys.path查找

For the final command, make sure the directory is in your default [Python][Python 1] path, which can be found with`sys.path` in a [python][Python 1] interpreter.

c.验证

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  import dlib

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  import dlib

### 7、Git获取openface ###

a.下载Openface

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git clone https://github.com/cmusatyalab/openface.git
2.  git submodule init 
3.  git submodule update

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  git clone https://github.com/cmusatyalab/openface.git
2.  git submodule init 
3.  git submodule update

b.在Openface根目录执行

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  sudo python2 setup.py install

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  sudo python2 setup.py install

python2一定要确保dlib和[OpenCV][]安装成功

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  import cv2 
2.  import dlib

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  import cv2 
2.  import dlib

c.获取模型

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  models/get-models.sh

**\[cpp\]** [view plain][] [copy][view plain]

[![在CODE上查看代码片][CODE]][CODE_CODE] [![派生到我的代码片][ico_fork.svg]][ico_fork.svg 1]

1.  models/get-models.sh

### 8、运行demo ###

运行demo1：

./demos/compare.pyimages/examples/\{lennon\*,clapton\*\}

运行demo2：

./demos/classifier.py infermodels/openface/celeb-classifier.nn4.small2.v1.pkl ./images/examples/carell.jpg

运行demo3：

./demos/web/start-servers.sh

### 9、demo1可用浏览器 ###

可用浏览器：

360浏览器极速模式

火狐浏览器

搜狗浏览器高速模式

不可用浏览器：

Chrome谷歌浏览器（可能与浏览器更新有关系，getUserMedia()）

IE览器

[http_cmusatyalab.github.io_openface]: http://cmusatyalab.github.io/openface/
[FaceNet_ A Unified Embedding for Face Recognition and Clustering]: http://www.cv-foundation.org/openaccess/content_cvpr_2015/app/1A_089.pdf
[Brandon Amos]: http://bamos.github.io/
[https_github.com_cmusatyalab_openface]: https://github.com/cmusatyalab/openface
[Link 1]: http://lib.csdn.net/base/operatingsystem
[http_cmusatyalab.github.io_openface_setup]: http://cmusatyalab.github.io/openface/setup/
[Link 2]: http://lib.csdn.net/base/softwaretest
[http_blog.csdn.net_itas109_article_details_50679251]: http://blog.csdn.net/itas109/article/details/50679251
[view plain]: http://blog.csdn.net/ren1335654481/article/details/54563906#
[CODE]: https://code.csdn.net/assets/CODE_ico.png
[CODE_CODE]: https://code.csdn.net/snippets/1596253
[ico_fork.svg]: https://code.csdn.net/assets/ico_fork.svg
[ico_fork.svg 1]: https://code.csdn.net/snippets/1596253/fork
[dpnn]: https://github.com/nicholas-leonard/dpnn
[nn]: https://github.com/torch/nn
[csvigo]: https://github.com/clementfarabet/lua---csv
[cunn]: https://github.com/torch/cunn
[fblualib]: https://github.com/facebook/fblualib
[DNN]: http://cmusatyalab.github.io/openface/training-new-models/
[torchx]: https://github.com/nicholas-leonard/torchx
[Git]: http://lib.csdn.net/base/git
[OpenCV]: http://lib.csdn.net/base/opencv
[https_github.com_Itseez_opencv_archive_2.4.11.zip]: https://github.com/Itseez/opencv/archive/2.4.11.zip
[http_docs.opencv.org_2.4_doc_tutorials_introduction_linux_install_linux_install.html]: http://docs.opencv.org/2.4/doc/tutorials/introduction/linux_install/linux_install.html
[https_github.com_davisking_dlib_releases_download_v18.16_dlib-18.16.tar.bz2]: https://github.com/davisking/dlib/releases/download/v18.16/dlib-18.16.tar.bz2
[Python]: http://lib.csdn.net/base/11
[Python 1]: http://lib.csdn.net/base/python