Skip to main content

Docker Installation

For maximum portability, the docker installer allows EyesOnIt to be installed on Windows, Linux or Mac computers. An NVIDIA GPU is required.

Installing the EyesOnIt Docker Image

Obtain Your License​

If you haven't done so already, email your EyesOnIt contact or info@eyesonit.us to obtain an account and a demo license.

Install the Software​

1. Install Docker​

Install Docker from here.

You can install Docker directly on Linux. Ubuntu is the prefered host environment.

If your server is running Windows, create a Ubuntu image in WSL2 and install Docker on that Ubuntu image.

2. Install the NVIDIA Docker Container Toolkit​

In your Linux distribution install the NVIDIA Container Toolkit from here

Also make sure you have the latest NVIDIA drivers on your host system.

3. Install EyesOnIt​

  • Pull our Docker image

    docker pull eyesonit/eyesonit_v4.0:260406.4

Note: you can find the latest release at https://hub.docker.com/orgs/eyesonit/repositories

Run the software​

Use this Docker run command to create a Docker container with GPU support:

docker run 
	-d 
	--label "your label"
	--gpus all   
	--mount type=volume,source=<your source>,target="/home/eyesonit_user/output"
	--mount type=tmpfs,target=/tmp,tmpfs-size=128m   
	--cap-drop=ALL 
	--security-opt no-new-privileges   
	-p 8000:8000   
	--restart unless-stopped   
	-e EOI_LICENSE_KEY="<your EyesOnIt license key>"   
	-e EOI_AUTHORIZATION_TOKEN="<your EyesOnIt license token>"   
	-e TZ="Country/City"
	-v "host input path":"/home/eyesonit_user/input" 
	"repository:tag"

Parameter notes:
--label "your label": create a label that has meaning to you
--mount type=volume,source=<your source>,target="/home/eyesonit_user/output": if your mount source is eyesonit_output, EyesOnIt output will appear in /var/lib/docker/volumes/eyesonit_output/_data. Don't change "/home/eyesonit_user/output"
-e TZ="Country/City": this parameter is optional. If included, provide your timezone
-v "host input path":"/home/eyesonit_user/input": "host input path" needs to be a path on your host that the container can read. Don't change "/home/eyesonit_user/input"
"repository:tag": replace this with the docker repository:tag that you pulled
Example docker run command​

Here's an example docker run command:

docker run -d --label "eyesonit/v4.0"
--gpus all
--mount type=volume,source=eyesonit_output,target="/home/eyesonit_user/output"
--mount type=tmpfs,target=/tmp,tmpfs-size=128m
--cap-drop=ALL --security-opt no-new-privileges
-p 8000:8000
--restart unless-stopped
-e EOI_LICENSE_KEY="XXXXXX-XXXXXX-XXXXXX-XXXXXX-XXXXXX-V3"
-e EOI_AUTHORIZATION_TOKEN="activ-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxv3"
-e TZ="America/Denver"
-v "/home/user_name/input":"/home/eyesonit_user/input"
"eyesonit/eyesonit_v4.0:2026-04-06.4"

Note: If you are receiving your RTSP stream through a video management system, you may have to open another port on the container. You can do that by adding something like "-p 654:654" after "-p 8000:8000".

Open the Web UI​

On the same computer where you ran EyesOnIt, open a browser and navigate to http://localhost:8000/dashboard. If that page opens successfully then EyesOnIt is running.

You can find more information about the test UI here