Skip to content

Latest commit

 

History

History
116 lines (76 loc) · 7.76 KB

installation.md

File metadata and controls

116 lines (76 loc) · 7.76 KB

Technical Considerations

For reference, we use e.g. Dell workstations (79xx series) with Ubuntu 16.04 LTS and run a Docker container that has TensorFlow, etc. installed (https://github.com/AlexEMG/Docker4DeepLabCut). The code also runs on Windows (thanks to Richard Warren for checking it) or MacOS (some users have already successfully done so).

  • Camera Hardware:

    • The software is very robust to track data from pretty much any camera (grayscale, color, or graysale captured under infrared light etc.). See demos on our website

  • Computer Hardware:

    • Ideally, you will use a strong GPU with at least 8GB memory such as the NVIDIA GeForce 1080 Ti. A GPU is not necessary, but on a CPU the (training and evaluation) code is considerably slower (100x). You might also consider using cloud computing services like Google cloud/amazon web services.

  • Software:

    • Linux: use only Ubuntu 16.xx and not 18.xx for ease of installation!
    • We highly recommend using the supplied Docker container. NOTE: this container does not work on windows hosts!
    • The toolbox is written in Python 3. You will need TensorFlow (we used version 1.0 in the paper, later versions also work with the provided code (we tested TensorFlow versions 1.0 to 1.4, 1.8, and 1.10) for Python 3 with GPU support.
    • To note, is it possible to run DeepLabCut on your CPU, but it will be VERY slow. However, this is the preferred path if you want to test DeepLabCut on your data before purchasing a GPU, with the added benefit of a straightforward installation.
    • WINDOWS specific note: If you want to run the commands in the program "cmd" (terminal), we recommend using Git Bash: https://git-scm.com/book/en/v2/Getting-Started-Installing-Git (You can use "terminal" in Linux/MacOS as is).

Anaconda environment with TensorFlow on GPU support:

  • Docker: We highly recommend to use the supplied Docker container, which has everything including TensorFlow for the GPU preinstalled! NOTE: this container does not work on windows hosts!

  • If you cannot use the Docker container, Anaconda is perhaps the easiest way to install Python and additional packages across various operating systems. First create an Anaconda environment with all dependencies, but TensorFlow. With Anaconda you create all the dependencies in an environment on your machine by running in a terminal:

LINUX/MacOS:

git clone https://github.com/AlexEMG/DeepLabCut.git
cd deeplabcut/conda-files
conda env create -f dlcdependencies.yml

Windows:

git clone https://github.com/AlexEMG/DeepLabCut.git
cd deeplabcut/conda-files
conda env create -f dlcDependenciesFORWINDOWS.yaml

Then, for LINUX & Windows: Install TensorFlow. We used TensorFlow 1.0 with CUDA (Cuda 8.0). Therefore, we recommend this version. Some other versions of TensorFlow have been tested, but use at your own risk (i.e. these versions have been tested 1.2, 1.4, 1.8 or 1.10, but might require different CUDA versions! Please check your driver/CUDA/TensorFlow version on this Stackoverflow post. Please check your CUDA and TensorFlow installation with the line below

  $ sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))

You can test that your GPU is being properly engaged with these additional tips.

Anaconda environment with TensorFlow for only a CPU:

Anaconda is perhaps the easiest way to install Python and additional packages across various operating systems. With Anaconda just create all the dependencies in an environment on your machine by running in a terminal:

LINUX/MacOS:

git clone https://github.com/AlexEMG/DeepLabCut.git
cd deeplabcut/conda-files
conda env create -f dlcdependencieswTF1.2.yml

Some conda channels are different for Windows and Rick Warren contributed the following yml file for Windows which actually features TensorFlow 1.8:

Windows:

git clone https://github.com/AlexEMG/DeepLabCut.git
cd deeplabcut/conda-files
conda env create -f dlcDependenciesFORWINDOWSwTF.yaml

That's it! You should now be able to run the .py files from inside the environment in the terminal.

Note that these environment yaml file might not work across platforms (other versions of Windows, MacOS, ...). Installing TensorTlow on your CPU is easy on any platform, if the environment yaml with TensorFlow does not work for you, then install the one without, and then follow the instructions here for installing Tensorflow 1.2.

Once you installed the environment you can activate it, by typing on Linux/MacOS:

source activate DLCdependencies

and on Windows do:

activate DLCdependencies

Then you can work with all DLC functionalities inside this environment. Enjoy!

Alternative Python installation (not Anaconda environment based):

If you wish to just install everything directly onto your computer (i.e. you don't need other versions of TF, CUDA, etc), then you can install these packages using the terminal:

  • Install Python3 [e.g. Sypder (or any other IDE) and/or Jupyter Notebook for Python3]

  • Clone (or download) the code we provide

    $ git clone https://github.com/AlexEMG/DeepLabCut

  • You will also need to install the following Python3 packages (in the terminal type):

      $ pip install scipy scikit-image sk-video matplotlib pyyaml easydict 
      $ pip install moviepy imageio tqdm tables
      $ pip install pyqt5 scikit-learn pandas
      $ pip install ffmpeg-python

Note if you also have python2 installed, you might want to use pip3 instead (DeepLabCut is suported in Python3.x only).

Then, install TensorFlow:

Then, for LINUX & Windows: Install TensorFlow. We used TensorFlow 1.0 with CUDA (Cuda 8.0). Therefore, we recommend this version. Some other versions of TensorFlow have been tested, but use at your own risk (i.e. these versions have been tested 1.2, 1.4, 1.8 or 1.10, but might require different CUDA versions! Please check your driver/CUDA/TensorFlow version on this Stackoverflow post. Here is a nice guide for CUDA + cuDNN for TensorFlow installation on Ubuntu.

Please check your CUDA and TensorFlow installation with th line below

  $ sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))

You can test that your GPU is being properly engaged with these additional tips.

Now you can use Jupyer Notebooks, Spyder, and to train just use the terminal, to run all the code!

Return to readme.