2019/01/29 21:19
阅读数 1.2K

Installing Faceswap


Machine learning essentially involves a ton of trial and error. You're letting a program try millions of different settings to land on an algorithm that sort of does what you want it to do. This process is really really slow unless you have the hardware required to speed this up.


The type of computations that the process does are well suited for graphics cards, rather than regular processors. It is pretty much required that you run the training process on a desktop or server capable GPU. Running this on your CPU means it can take weeks to train your model, compared to several hours on a GPU.


Hardware Requirements硬件需求

TL;DR: you need at least one of the following:你需要以下至少一个:

  • A powerful CPU强大的CPU
    • Laptop CPUs can often run the software, but will not be fast enough to train at reasonable speeds电脑CPU通常可以运行这个软件,但是可能不够快。
  • A powerful GPU强大的GPU
    • Currently only Nvidia GPUs are supported. AMD graphics cards are not supported. This is not something that we have control over. It is a requirement of the Tensorflow library.目前只有Nvidia GPU支持。AMD显卡不支持。这不是我们能够改变的。因为它需要用到Tensorflow库。
    • The GPU needs to support at least CUDA Compute Capability 3.0 or higher. To see which version your GPU supports, consult this list: https://developer.nvidia.com/cuda-gpus Desktop cards later than the 7xx series are most likely supported.GPU需要支持至少CUDA Compute Capability 3.0以上。要看你的GPU支持哪个版本,查询这个列表:https://developer.nvidia.com/cuda-gpus 比7xx系列晚的Destop卡最可能支持。
  • A lot of patience十足的耐心。

Supported operating systems支持的操作系统

  • Windows 10 Windows 7 and 8 might work. Your milage may vary。Win10 Win7 Win8可能可以工作。但是可能麻烦一些。
  • Linux Most Ubuntu/Debian or CentOS based Linux distributions will work. Linux大多数基于Ubuntu/Debian或CentO S的Linux分支可以工作。
  • macOS GPU support on macOS is limited due to lack of drivers/libraries from Nvidia. 由于缺少Nvidia的驱动程序/库,MacOS对MacOS的GPU支持有限

Alternatively there is a docker image that is based on Debian.

Important before you proceed

In its current iteration, the project relies heavily on the use of the command line, although a gui is available. if you are unfamiliar with command line tools, you may have difficulty setting up the environment and should perhaps not attempt any of the steps described in this guide. This guide assumes you have intermediate knowledge of the command line.

The developers are also not responsible for any damage you might cause to your own computer.

General Install Guide

Installing dependencies

  • Python >= 3.2

    • apt/yum install python3 (Linux)
    • Installer (Windows)
    • brew install python3 (macOS)
  • virtualenv and virtualenvwrapper may help when you are not using docker.

  • If you are using an Nvidia graphics card You should install CUDA (https://developer.nvidia.com/cuda-zone) and CUDNN (https://developer.nvidia.com/cudnn). If you do not plan to build Tensorflow yourself, make sure you install no higher than version 9.0 of CUDA and 7.0.x of CUDNN

  • dlib is required for face recognition and is compiled as part of the setup process. You will need the following applications for your os to successfully install dlib (nb: list may be incomplete. Please raise an issue if another prerequisite is required for your OS):

    • Windows: Visual Studio 2015, CMake v3.8.2
    • Linux: build-essential, cmake
    • macOS: xquartz

Getting the faceswap code

Simply download the code from http://github.com/deepfakes/faceswap - For development it is recommended to use git instead of downloading the code and extracting it.

For now, extract the code to a directory where you're comfortable working with it. Navigate to it with the command line. For our example we will use ~/faceswap/ as our project directory.


Enter the folder that faceswap has been downloaded to and run:

python setup.py

If setup fails for any reason you can still manually install the packages listed within requirements.txt

About some of the options

  • CUDA: For acceleration. Requires a good nVidia Graphics Card (which supports CUDA inside)
  • Docker: Provide a ready-made image. Hide trivial details. Get you straight to the project.
  • nVidia-Docker: Access to the nVidia GPU on host machine from inside container.

CUDA with Docker in 20 minutes.

INFO    The tool provides tips for installation
        and installs required python packages
INFO    Setup in Linux 4.14.39-1-MANJARO
INFO    Installed Python: 3.6.5 64bit
INFO    Installed PIP: 10.0.1
Enable  Docker? [Y/n] 
INFO    Docker Enabled
Enable  CUDA? [Y/n] 
INFO    CUDA Enabled
INFO    1. Install Docker
        2. Install Nvidia-Docker & Restart Docker Service
        3. Build Docker Image For Faceswap
        docker build -t deepfakes-gpu -f Dockerfile.gpu .
        4. Mount faceswap volume and Run it
        # without gui. tools.py gui not working.
        nvidia-docker run --rm -it -p 8888:8888 \
            --hostname faceswap-gpu --name faceswap-gpu \
            -v /opt/faceswap:/srv \
        # with gui. tools.py gui working.
        ## enable local access to X11 server
        xhost +local:
        ## enable nvidia device if working under bumblebee
        echo ON > /proc/acpi/bbswitch
        ## create container 
        nvidia-docker run -p 8888:8888 \
            --hostname faceswap-gpu --name faceswap-gpu \
            -v /opt/faceswap:/srv \
            -v /tmp/.X11-unix:/tmp/.X11-unix \
            -e DISPLAY=unix$DISPLAY \
            -e AUDIO_GID=`getent group audio | cut -d: -f3` \
            -e VIDEO_GID=`getent group video | cut -d: -f3` \
            -e GID=`id -g` \
            -e UID=`id -u` \
        5. Open a new terminal to interact with the project
        docker exec faceswap-gpu python /srv/tools.py gui

A successful setup log, without docker.

INFO    The tool provides tips for installation
        and installs required python packages
INFO    Setup in Linux 4.14.39-1-MANJARO
INFO    Installed Python: 3.6.5 64bit
INFO    Installed PIP: 10.0.1
Enable  Docker? [Y/n] n
INFO    Docker Disabled
Enable  CUDA? [Y/n] 
INFO    CUDA Enabled
INFO    CUDA version: 9.1
INFO    cuDNN version: 7
WARNING Tensorflow has no official prebuild for CUDA 9.1 currently.
        To continue, You have to build your own tensorflow-gpu.
        Help: https://www.tensorflow.org/install/install_sources
Are System Dependencies met? [y/N] y
INFO    Installing Missing Python Packages...
INFO    Installing tensorflow-gpu
INFO    Installing pathlib==1.0.1
INFO    Installing tqdm
INFO    Installing matplotlib
INFO    All python3 dependencies are met.
        You are good to go.

Run the project

Once all these requirements are installed, you can attempt to run the faceswap tools. Use the -h or --help options for a list of options.

python faceswap.py -h

or run with gui to launch the GUI

python faceswap.py gui

Proceed to ../blob/master/USAGE.md


This guide is far from complete. Functionality may change over time, and new dependencies are added and removed as time goes on.

If you are experiencing issues, please raise them in the faceswap-playground repository instead of the main repo.

Windows Install Guide

Setting up Faceswap can seem a little intimidating to new users, but it isn't that complicated, although a little time consuming. It is recommended to use Linux where possible as Windows will hog about 20% of your GPU Memory, making Faceswap run a little slower, however using Windows is perfectly fine and 100% supported.


Microsoft Visual Studio 2015

Important Make sure to downoad the 2015 version of Microsoft Visual Studio

Download and install Microsoft Visual Studio 2015 from: https://go.microsoft.com/fwlink/?LinkId=532606&clcid=0x409

On the install screen:

  • Select "Custom" then click "Next"
    MSVS Custom
  • Uncheck all previously checked options
  • Expand "Programming Languages" and select "Visual C++"
    MSVS C++
  • Select "Next" and "Install"


GPU Only If you do not have an Nvidia GPU you can skip this step.

At the time of writing Tensorflow (version 1.12) only supports Cuda up to version 9.0, but check https://www.tensorflow.org/install/gpu for the latest supported version. It is crucial that you download the correct version of Cuda.

Download and install the correct version of the Cuda Toolkit from: https://developer.nvidia.com/cuda-toolkit-archive

NB: Make a note of the install folder as you'll need to access it in the next step.


GPU Only If you do not have an Nvidia GPU you can skip this step.

As with Cuda you will need to install the correct version of cuDNN that the latest Tensorflow supports. At the time of writing this is Tensorflow v1.12 which supports cuDNN version 7.2, but check https://www.tensorflow.org/install/gpu for the latest supported version.

Download cuDNN from https://developer.nvidia.com/cudnn. You will need to create an account with Nvidia.

At the bottom of the list of latest cuDNN release will be a link to "Archived cuDNN Releases": cuDNN Archive

Select this and choose the latest version of cuDNN that supports the version of Cuda you installed and has a minor version greater than or equal to the latest version that Tensorflow supports. (Eg Tensorflow 1.12 supports Cuda 9.0 and cuDNN 7.2. There is not an archived version of cuDNN 7.2 for Cuda 9.0, so select cuDNN version 7.3)

  • Open the zip file
  • Extract all of the files and folders into your Cuda folder (It is likely to be located in C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA):
    cuDNN to Cuda


Install the latest stable release of CMake from https://cmake.org/download/. (Scroll down the page for Latest Releases and select the relevant Binary distribution installer for your OS).

When installing CMake make sure to enable the option to CMake to the system path: cmake path


Download and install the latest Python 3 Anacconda from: https://www.anaconda.com/download/. Unless you know what you are doing, you can leave all the options at default.


Download and install Git for Windows: https://git-scm.com/download/win. Unless you know what you are doing, you can leave all the options at default.


Reboot your PC, so that everything you have just installed gets registered.


Set up a virtual environment

  • Open up Anaconda Navigator
  • Select "Environments" on the left hand side
  • Select "Create" at the bottom
  • In the pop up:
    • Give it the name: faceswap
    • IMPORTANT: Select python version 3.5
    • Hit "Create" (NB: This may take a while as it will need to download Python 3.5) Anaconda virtual env setup

Entering your virtual environment

To enter the virtual environment:

  • Open up Anaconda Navigator
  • Select "Environments" on the left hand side
  • Hit the ">" arrow next to your faceswap environment and select "Open Terminal" Anaconda enter virtual env


  • If you are not already in your virtual environment follow these steps
  • Get the Faceswap repo by typing: git clone https://github.com/deepfakes/faceswap.git
  • Enter the faceswap folder: cd faceswap

Easy install

  • Enter python setup.py and follow the prompts.

If you have issues/errors follow the Manual install steps below.

Manual install

If dlib failed to install you can follow the steps to manually install dlib.
Once dlib is installed follow these steps:

  • Install tkinter (required for the GUI) by typing: conda install tk
  • Install requirements: pip install -r requirements.txt
  • Install Tensorflow (either GPU or CPU version depending on your setup):
    • GPU Version: pip install tensorflow-gpu
    • Non GPU Version: pip install tensorflow

Running Faceswap

  • If you are not already in your virtual environment follow these steps
  • Enter the faceswap folder: cd faceswap
  • Enter the following to see the list of commands: python faceswap.py -h or enter python faceswap.py gui to launch the GUI

Create a desktop shortcut

A desktop shortcut can be added to easily launch staight into the faceswap GUI:

  • Open Notepad
  • Paste the following:
%USERPROFILE%\Anaconda3\envs\faceswap\python.exe %USERPROFILE%/faceswap/faceswap.py gui
  • Save the file to your desktop as "faceswap.bat"

Updating faceswap

It's good to keep faceswap up to date as new features are added and bugs are fixed. To do so:

  • If you are not already in your virtual environment follow these steps
  • Enter the faceswap folder: cd faceswap
  • Enter the following git pull --all
  • Once the latest version has downloaded, make sure your requirements are up to date: pip install --upgrade -r requirements.txt


You should only need to follow these steps if you want the latest Dlib code or the process was unable to install Dlib for you.

For reasons outside of our control, this is the trickiest part of the process, and most of the prerequisites you installed are to support just Dlib. It is recommended to build Dlib from source for 3 main reasons:

  1. To get the latest version
  2. Enable GPU Support in Dlib
  3. To prevent yourself running into a whole host of issues later in the process.

If you are not bothered about having GPU support or the latest version, scroll to the end of this section for a simple one-liner to install the CPU version of Dlib.

Build Latest Dlib with GPU Support

  • If you are not already in your virtual environment follow these steps
  • In the terminal type: git clone https://github.com/davisking/dlib.git
  • Enter the dlib folder: cd dlib
  • Add Visual Studio to your path by typing: SET PATH=%PATH%;C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\bin
  • Enter: python setup.py -G "Visual Studio 14 2015" install --yes USE_AVX_INSTRUCTIONS --yes DLIB_USE_CUDA --clean

This will build and install dlib for you. It is worth backing up the generated .egg file somewhere so that you can re-install it if you ever need to rather than having to re-compile:

  • From within the dlib folder copy the file named dlib-xx.yy.zz-py3.5-win-amd64.egg to somewhere safe
  • If you ever need to install it again, then from within your virtual environment enter: python -m easy_install <path to saved .egg>

Once Dlib is built, you can remove Visual Studio and CMake from your PC.

Easy install of Dlib without GPU Support

NB: Don't do this if you have already compiled Dlib with GPU support.

  • If you are not already in your virtual environment follow these steps
  • In the terminal type: conda install -c conda-forge dlib
0 收藏
0 评论
0 收藏