If you just want to install encord-agents in your current environment, you can run:

python -m pip install encord-agents

Optionally include additional dependencies used for working with visual files such as Images, Videos, Image Groups / Sequences by running:

python -m pip install "encord-agents[vision]"

Installing the vision extras is advised for most users. These dependencies optimize performance when handling visual data formats.

If you are building serverless deployments (for example: AWS Lambda) or working with non-visual data (labels, priorities, text, PDFs, or audio), you may skip the vision extras to minimize bundle size.

Encord Agents requires python >= 3.10. If you do not have Python 3.10, we recommend usingpyenv to manage your Python versions.

We recommend installing the Encord agents library in an isolated environment. There are several options available.

​
Venv

  1. Create a new virtual environment. In this example, we name it agents-venv but you can choose any name you see fit.
python -m venv agents-venv
  1. Next, activate your virtual environment. This ensures that any modules you install remain isolated and that all previously installed modules in the environment are available for use in Python.
source agents-venv/bin/activate
  1. Once activated, the environment name is displayed before the cursor in your terminal.
example
(agents-venv) $
  1. Install the encord-agents library.
python -m pip install encord-agents

​
Poetry

If you already have a Poetry project, you can add encord-agents to that project:

poetry add encord-agents

​
Conda

  1. Create a conda environment.
conda create -n agents python>=3.10
  1. We suggest activating your conda environment.
conda activate agents
  1. Once activated, the environment name is displayed before the cursor in your terminal.
example
(agents) $
  1. Install the encord-agents library.
python -m pip install encord-agents

​
Dependencies

The dependencies for encord-agents are kept lightweight, with the only relatively large dependencies being opencv-python and numpy. To see the full list of dependencies, see here.