QGeoAI logoQGeoAI

Installation

QGeoAI Tools is a suite of QGIS plugins powered by a local server. The server acts as the engine that handles all AI computations (PyTorch, SAM2, YOLO), while the plugins provide the interface within QGIS.

1. QGeoAI Server (Required)

QGeoAI Server is a local server that enables QGIS plugins (QAnnotate, QModel Trainer, QPredict, QToolbox) to access AI tools (PyTorch, SAM2, YOLO) without burdening QGIS.

Architecture Benefits

  • Dependency isolation (PyTorch, Ultralytics)
  • No model reloading between operations
  • QGIS stays lightweight and responsive
  • Automatic CUDA support for NVIDIA GPUs

Privacy & Security

  • 100% local (127.0.0.1 only)
  • No external connections
  • No telemetry or tracking
  • Token-based authentication

Prerequisites

  • Python 3.10 or higher
  • 5 GB of disk space (for SAM2 models)
  • NVIDIA GPU with CUDA support (optional, but recommended for performance)

Download

Contact us

Installation

# Windows python -m venv %USERPROFILE%\.qgeoai\env
# Linux / Mac python3 -m venv ~/.qgeoai/env
# Install server cd qgeoai_server python install_server.py

The installation script will: copy files to ~/.qgeoai/server/, install dependencies (FastAPI, PyTorch, etc.), download SAM2 models (~1.5 GB), detect and configure GPU if available, and create startup scripts.

Installation time: 10-20 minutes

Verify Installation

python check_installation.py

YOLO11 Models (Optional)

To use YOLO11 with pretrained weights:

  1. Download models from Ultralytics documentation
    https://docs.ultralytics.com/models/yolo11/
  2. Place .pt files in the models directory
    # Windows %USERPROFILE%\.qgeoai\server\models\ # Linux / Mac ~/.qgeoai/server/models/
  3. Compatible models: Detection (yolo11n.pt, yolo11s.pt, yolo11m.pt, yolo11l.pt, yolo11x.pt), Segmentation (yolo11n-seg.pt, yolo11s-seg.pt, etc.), OBB (yolo11n-obb.pt, yolo11s-obb.pt, etc.)
QGeoAI - GEOAI Tools for QGIS