MustatilCore

MustatilCore is the public Python API package for the Mustatil geospatial AI backend. It exposes generated wrappers for detection, training, geospatial processing, export, model utilities, pipelines, runtime helpers and job polling. The package is backend-focused and does not launch the desktop GUI. Heavy AI/runtime dependencies such as PyTorch, Ultralytics, Rasterio, GeoPandas, Fiona, PyProj and model-specific packages should be installed in the target environment according to the user's hardware and workflow.

Included capabilities

• YOLO detection and training helpers
• RF-DETR detection and training helpers
• DINO / Grounding DINO / LAE-DINO backend functions
• OWL / OWLv2 integration helpers
• SAM / SAM2 segmentation helpers
• Faster R-CNN, Mask R-CNN and U-Net backend functions
• GeoTIFF, raster, tiling, crop and preview utilities
• GPKG, GeoJSON, CSV and feature export helpers
• GeoNMS, filtering and FormLearner-related utilities
• Visual/AI pipeline backend functions
• Video detection and annotation-related helpers
• Runtime dependency checks, model utilities and job polling helpers

Public API coverage

Generated raw public wrappers: 15,625 The complete generated function list is included in FULL_FUNCTION_INDEX.md, FUNCTION_INDEX_SUMMARY.json, and is also appended at the end of this README.

Main function groups

Installation

pip install mustatilcore

Quick start

import mustatilcore
from mustatilcore import functions

print(functions.RAW_FUNCTION_COUNT)
print(len(functions.__all__))

# Friendly alias for first occurrence, when available:
# mustatilcore.detect_yolo(...)

# Full raw wrappers are named f_00001__module__function etc.

CLI

mustatilcore-functions
mustatilcore-describe detect_yolo
mustatilcore-call detect_yolo --args '["image.tif", ["model.pt"]]' --kwargs '{"device":"cuda"}'

Complete function list

Total raw public wrappers: 15,625