Installation

To install Pixano you need a working Python installation. It currently requires Python 3.10 to 3.13. We encourage installing Pixano in a virtual environment, or using the Docker distribution, which needs no Python at all. Pixano is supported on Linux and macOS, and on Windows with limited support.

Install with pip

Open a terminal and check your Python version first:

python3 --version

If it is not between 3.10 and 3.13 (Python 3.14 is not supported yet), install a supported version from python.org and use its own command below, for example python3.13 instead of python3. Do not skip this check: with an unsupported Python, pip installs a very old Pixano without any warning.

From the folder where you want to keep your Pixano files, create a virtual environment and activate it:

python3 -m venv .venv
source .venv/bin/activate

On Windows, use python instead of python3, and activate with .venv\Scripts\activate.

Install with:

pip install pixano

Hugging Face Hub and LeRobot imports are included in this installation. When running from a source checkout, uv sync installs their dependencies as well.

Video and LeRobot frame imports also need FFmpeg, including both ffmpeg and ffprobe, on your system’s PATH. Install it with your system package manager:

# Ubuntu / Debian
sudo apt-get install ffmpeg

# macOS with Homebrew
brew install ffmpeg

On Windows, install FFmpeg and add its bin directory to PATH. Verify the installation with ffmpeg -version and ffprobe -version. The Pixano Docker image already includes these tools. Image-only imports do not need them.

Then check it is correctly installed with:

pixano --help

 Usage: pixano [OPTIONS] COMMAND [ARGS]...

 Pixano CLI

The list of commands that follows includes init and server, the two you need next.

At this point you can start the Pixano server, by first initializing its data directory — the folder where Pixano keeps your datasets:

pixano init ./my_data

and then running the server itself:

pixano server run ./my_data

Open http://127.0.0.1:7492 in your browser: this is the Library, Pixano’s home page, empty for now. Continue to the Quickstart to import your first dataset.

The server runs until you press Ctrl-C in the terminal. To start it again later, open a terminal in the same folder, activate the virtual environment and run pixano server run ./my_data — pixano init is only needed once.

Run with Docker

With Docker installed and running, pull the image and start a container:

docker pull pixano/pixano:stable

docker run -d --name pixano -e DATA_DIR=/library -p 7492:7492 -v "$(pwd)/my_data:/library" pixano/pixano:stable

my_data is the folder where Pixano keeps your datasets; it is created in the current folder on first start. Open http://localhost:7492 in your browser: the Library.

Stop the container with docker stop pixano and start it again with docker start pixano.