Prepare your server
A little preparation makes setup straightforward. Choose your server, decide how your team will reach it, and have your installation details ready.
This guide prepares the server that runs EyesOnIt. Your team will use the EyesOnIt web UI from their browser computers.
Choose the installation type
Section titled “Choose the installation type”| Installation | Supported server operating systems | What the installer sets up |
|---|---|---|
| Native | Ubuntu 24.04, x86-64 | EyesOnIt services, analysis runtime, models, and HTTPS. |
| Docker | Ubuntu 24.04, Debian 13, Fedora 43 or 44, x86-64 | Docker Engine, Compose, NVIDIA Container Toolkit, and the EyesOnIt application. |
The installation walkthrough uses the native Linux download. For Docker, obtain the Docker installer from your EyesOnIt contact; its guided setup asks for the same site and account details.
Plan processing and recording capacity
Section titled “Plan processing and recording capacity”- Choose an NVIDIA GPU supported by the EyesOnIt TensorRT runtime, with compute capability 7.5 or later. Confirm the GPU model and workload with your EyesOnIt contact when sizing a server.
- Plan CPU, RAM, and GPU memory around the camera count, video resolution, analyzed frame rate, and features you intend to use. Add cameras in stages and review resource usage as you expand.
- Provide space for the application, model downloads, and recordings. The installer checks its required capacity and operational reserve; allow additional room for the video history you want to retain.
- For Docker, prepare a working NVIDIA driver 575.57.08 or later before installation. Native setup can prepare the supported driver and runtime dependencies and will explain any required restart.
For a recording estimate, 1 Mb/s uses about 10.8 GB per day of continuous video, before additional storage overhead. Multiply by the cameras’ recording bitrates and retention days. Storage and retention explains how to manage that history after setup.
Prepare network access
Section titled “Prepare network access”- Choose a stable server IP address or hostname for your team to use. If using a hostname, configure DNS to point to the server or your HTTPS proxy.
- Allow browser access to EyesOnIt over HTTPS, port 443, on the selected server address, or through your existing proxy.
- Confirm that the server can reach your cameras’ video addresses and ports. A browser computer’s camera access does not establish access from the server.
- Provide outbound access to the EyesOnIt download and licensing services and the operating-system and NVIDIA package repositories. Docker setup also needs access to the Docker repositories and image registry.
Use an account with sudo access to run the installer. Keep the server connected to power and the network while it downloads and prepares the analysis models.
Choose HTTPS and gather your details
Section titled “Choose HTTPS and gather your details”| Have ready | What it is used for |
|---|---|
| EyesOnIt license key and authorization token | Activate the licensed installation. Enter them in the installer’s masked prompts. |
| Browser address | The hostname or IP your team will use to open EyesOnIt. |
| HTTPS choice | Let EyesOnIt create a local certificate, supply an existing certificate and private key, or use an existing HTTPS reverse proxy. |
| Administrator details | A username, display name, optional email, and a private password for the initial administrator. |
| First camera details | Its stream address, credentials, location, and time zone. These are used after installation. |
For a local certificate, plan to trust the EyesOnIt public CA certificate on each browser computer. For an existing certificate, have the PEM certificate-chain and matching private-key files available on the server; the certificate must cover your chosen browser address. If using a proxy, have your administrator configure HTTPS and forwarding to EyesOnIt on the same server at 127.0.0.1:8080.
Continue with Install EyesOnIt.