DeepLabCut Installation
(Credit: Nick Morrison, Unsplash.com)
We are going to follow the installation instructions that are given on the installation documentation page.
Download the miniconda suite
We use the recommended miniconda installation. You can download the appropriate miniconda installer from the Anaconda site. Select the platform: Intel or Apple M1. You can do a graphical installation (pkg) or via terminal (bash). The pkg installation is easier.
We start installing PyTorch
Step 1: Create a new conda environment
conda create --name deeplabcut python=3.10
Step 2: Activate the conda enviroment
conda activate deeplabcut
(This environment is created under /opt/miniconda3/envs, in case you want to undo everything.)
Step 3: Install PyTorch
Paste the below code in the terminal (make sure you’re in the env)
conda install pytorch::pytorch torchvision torchaudio -c pytorch
Reference: Pytorch for Mac M1/M2 with GPU acceleration 2023. Jupyter and VS Code setup for PyTorch included.. Mustafa Mujahid. Medium, 2023.
If you use MacOS/Linux, you can take advantage of the Homebrew software administration tool, to manage open source software.
If you haven't installed brew yet, run the following script from a terminal shell
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
Step 1: We will use brew to install
brew install hdf5
Step 2: Install the tensorflow package for macos if M1 chip.
pip install tensorflow-macos
and
Step 3: Install the GPU accelerator for Apple M1
pip install tensorflow-metal
Reference: TensorFlow with GPU support on Apple Silicon Mac with Homebrew and without Conda / Miniforge. Søren L Kristiansen. Medium, 2022.
We describe a series of steps to install DeepLabCut, with pytorch support.
Step 1: Install tensorpack
conda install hcc::tensorpack
Step 2: Install tf_slim
conda install conda-forge::tf_slim
Step 3: Install pytables
conda install -c conda-forge pytables==3.8.0
Step 4: Install DeepLabCut source form Github
pip install "git+https://github.com/DeepLabCut/DeepLabCut.git@pytorch_dlc#egg=deeplabcut[gui,modelzoo,wandb]"
If there are no missing packages, then you are ready to run deeplabcut
python -m deeplabcut
Then you will get the active GUI
You are ready to start with your projects.
Use command Q in the MacBook to finish.
Created: 08/09/2024 (C. Lizárraga)
Updated: 08/15/2024 (C. Lizárraga)
DataLab, Data Science Institute, University of Arizona.
