mustatil 5.6.0

Mustatil GIS-level AI vision workspace for annotation, YOLO training, large-image detection and geospatial export.

Mustatil 5.6.0

Mustatil is a GIS-level AI vision workspace for annotation, YOLO training, large-image / GeoTIFF detection, satellite-map detection, visual AI pipelines, remote-sensing workflows, and GPKG/GIS export.

Install from PyPI

python -m pip install --upgrade pip
python -m pip install --upgrade mustatil
mustatil

Alternative start command:

python -m mustatil

Included in this PyPI build

• Qt/PySide6 desktop workspace
• Detection and Satellite Detection helpers
• Annotator, trainer, video, remote-sensing, GIS review, AI pipeline, RCNN/UNet/SAM2/RF-DETR/LAE-DINO helper plugins from this release
• Starter model/data files where PyPI size allows:
  • Houses300.pt
  • Houses300.onnx
  • Tree50.pt
  • yolov8n.pt
  • yolo26n.pt
  • sam2_all_polygons.json
• Python dependency declarations so pip install mustatil pulls the required Python packages.

Large model note

sam2_b.pt from the original folder is larger than PyPI's normal single-file upload limit, so it is not bundled in this PyPI-ready wheel. Place sam2_b.pt next to mustatil_qt_workspace.py in the installed package directory or select it manually in the UI if you want to use that exact local checkpoint.

GPU note

PyPI dependency metadata cannot force the CUDA-specific PyTorch wheel index. A plain pip install mustatil installs PyTorch through normal package resolution. For NVIDIA CUDA builds, install the matching PyTorch CUDA wheels first, then install or upgrade Mustatil.