[general]
name = Spatium Labs Vision
description = RGB image segmentation into editable land-cover polygons and reviewed training samples.
version = 0.5.2
qgisMinimumVersion = 3.28
author = Spatium Labs
category = Raster
experimental = True
hasProcessingProvider = False
license = GPL-2.0-or-later
about = Segments visible QGIS imagery or local georeferenced RGB rasters into candidate polygons. Requires a separate Python 3.10-3.12 environment with rasterio, shapely, pyproj, fiona, torch, transformers and onnxruntime; dependencies and model weights are downloaded on demand. No packages are installed into QGIS Python. Image rights and model terms apply. Outputs require manual review; no cadastral or crop identification guarantee. Turkish interface.
icon = icon.png
qgisMaximumVersion = 3.99
email = melih44523@gmail.com
deprecated = False
homepage = https://github.com/melihanileroglu/SpatiumLabs-QGIS/tree/main/spatium_landcover
repository = https://github.com/melihanileroglu/SpatiumLabs-QGIS
tracker = https://github.com/melihanileroglu/SpatiumLabs-QGIS/issues
tags = remote sensing,segmentation,land cover,buildings,RGB
changelog = 0.5.0: Initial public preview. Animated introduction, documented installation and model dependencies, validated packages, Network barriers and ring outputs, Vision multiclass polygons and topology-aware simplification.
