OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Autoptz

AutoPTZ is a cross-platform desktop application that detects and tracks people in camera video, then sends pan, tilt and zoom commands to a PTZ camera to keep a selected person framed. It supports multi-camera workflows, optional face recognition and appearance-based re-identification, and VISCA, ONVIF and NDI camera-control backends.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Autoptz — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use AutoPTZ when a live camera workflow needs automated framing around a chosen person and the camera can be controlled through a supported VISCA, ONVIF or NDI backend.

Who it is for

  • Live-production teams that need a camera to follow a selected presenter or participant.
  • Event operators setting up person-following across multiple cameras.
  • Developers or technically equipped users who can install desktop software and connect supported PTZ control interfaces.

Documented uses

  • Select a person in the camera view and have a PTZ camera follow them.
  • Maintain or recover the selected target’s identity after an occlusion.
  • Run independent tracking workers for multiple cameras.
  • Choose an ONNX Runtime execution option suited to the computer’s platform and available accelerator.
Before you choose

Requirements & limits

The project describes prebuilt macOS DMG, Windows EXE and Linux AppImage installers on its Releases page. Building from source requires Python 3.12 or later. The repository states that AutoPTZ is licensed under AGPL-3.0.

  • Confirm that the camera exposes a supported control route: VISCA over USB or IP, ONVIF, or NDI.
  • The desktop vision pipeline uses ONNX Runtime; available execution options include CoreML, TensorRT/CUDA, DirectML, OpenVINO and CPU.
  • Detection and pose features depend on models being cached; some controls remain disabled without required models.
  • Face recognition and appearance re-identification are optional features, not prerequisites for basic person tracking.
  • The README states that only one onnxruntime package wheel can be installed at a time.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install and prepare AutoPTZ from source

Set up the source checkout and prepare its detection, tracking and interface stack.

Before you start

  • Python 3.12 or later
  • A desktop operating system supported by the project

Steps

  1. Clone the AutoPTZ repository and change into its directory.

  2. Create a virtual environment at the project root and activate it.

  3. Run the command below to install the full stack.

    Documented command
    python tools/install.py --editable

  4. Launch the app with the command below.

    Documented command
    python -m autoptz

  5. Use Engine → Models... to cache detector tiers and the pose model.

Expected outcome

The app is launched from the source checkout, with model-dependent features available after the relevant models are cached.

Guide limitations

  • Without required models, affected feature controls are disabled; live preview remains available where possible.
  • Missing detector tiers are not fetched on switching unless automatic model downloads are enabled.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
AutoPTZ/autoptz · Official homepage title · retained in its source language
How we review sourcesSuggest a correction