Mustatil Legacy for Windows

Mustatil Legacy, developed by TWasfy, is a local desktop AI and GIS workspace for detecting and annotating large-scale imagery. It focuses on offline processing of satellite and aerial mosaics, enabling detection, annotation, and dataset preparation on a single machine without cloud services. The tool bundles annotation editing, detection previewing, and geospatial export options. It is offered as the legacy build, version 5.6, and targets researchers, archaeologists, and GIS professionals working with high-resolution imagery.

What tasks can you actually use it for?

Mustatil maps to large-raster object-detection and dataset-preparation work, using a tiled detection pipeline to break huge imagery into processable parts. The tool supports local dataset crop export and editable class management, so users can build training sets and refine labels inside the same desktop environment. Its detection preview lets operators inspect scored results before committing exports or creating training crops.

How reliable are the detection outputs for professional workflows?

The tool applies YOLO-based tiled detection, and it surfaces per-detection scores to help triage results. That design makes reliability dependent on chosen models and source imagery quality, since scoring and previewing are the main mechanisms for filtering false positives. Because annotation editing is built in, practitioners prepare and correct training data locally rather than relying solely on automated outputs.

What file formats and large-raster workflows does it support?

Mustatil handles very large geospatial rasters and satellite mosaics that commonly destabilize standard viewers. Supported export targets include:

• GeoPackage and GeoJSON for geospatial vector outputs
• CSV for tabular metadata
• Cropped images for training dataset assembly

Does it fit into a privacy-focused, desktop workflow?

The app uses an offline-first architecture so all processing, inference, and storage remain on the user’s Windows desktop. It is built for Windows 10 and Windowsbit) and integrates annotation, detection previewing, and dataset preparation in one interface. The developer emphasizes local processing and the project is noted within scientific and archaeological circles for handling sensitive, large-format imagery on-machine.

Practical choice for specialists who need an inspectable legacy baseline

Mustatil is a pragmatic option for specialists requiring an on-machine toolkit that produces inspectable detection outputs and local training crops. As a legacy build, it does not include newer segmentation or model-chaining workflows found in the current generation, so teams needing those pipelines should evaluate the later workspace. The build suits projects that prioritize reproducibility and local control over the newest processing features.