GeoOSAM
Plugin ID: 4007
Advanced geospatial segmentation tool powered by SAM 2.1, SAM3 and Ultralytics with multi-point prompting, text/similar object detection, multi-format export, GPU/CPU optimization, and professional GIS workflow features
(213) votes
Download latest
GeoOSAM brings Meta's Segment Anything Models (SAM 2.1 and SAM3) to QGIS with intelligent hardware optimization. Features include multi-point segmentation with Shift/Ctrl click accumulation for precise control, SAM3 text-prompted detection, similar object finding across entire rasters, and multi-format export (GeoPackage, Shapefile, GeoJSON, FlatGeobuf). Automatic model selection: SAM 2.1 for GPU, Ultralytics SAM2.1_B for CPU with 5-10x performance gains. Supports point, bounding box, and batch segmentation modes, 12 pre-defined classes plus custom classes, undo functionality, tiled processing for large rasters, and Pro licensing for full-raster text/similar modes. Optimized for NVIDIA GPU, Apple Silicon, and multi-core CPU with intelligent threading. Perfect for remote sensing, urban planning, environmental monitoring, and geospatial analysis.
python
remote sensing
gis
urban planning
machine learning
segmentation
deep learning
satellite imagery
ai
sam
computer vision
semantic segmentation
sam2.1
pytorch
cuda
performance optimization
geospatial analysis
sam2
apple silicon
environmental monitoring
raster analysis
gpu acceleration
object detection
intelligent model selection
ultralytics
sam2.1_b
