Annotate
Create datasets with object classes, positive examples, correction boxes and training-ready labels.
Train. Detect. Annotate. Export.
A dark desktop workflow for custom YOLO models, large-scale geospatial imagery, video detection, annotation and export-ready results.
YOLO Trainer combines dataset creation, model training, image/video inference, correction workflows and export tools in one desktop application.
YOLO Trainer is designed for users who want to train custom object-detection models without splitting the workflow across many separate tools. It supports annotation, YOLO training, detection in large images, video frame detection, manual review and result export for downstream GIS or data workflows.
Practical training and detection tools for images, videos, geospatial datasets and review/export workflows.
From raw images to trained models and exportable detections.
Create datasets with object classes, positive examples, correction boxes and training-ready labels.
Train custom YOLO models using your own image folders, labels, class definitions and training settings.
Run inference on large images or videos, with confidence filtering, tiling and optional GPU acceleration.
Save reviewed results as GIS-ready and analysis-ready outputs, including GeoJSON, GPKG and CSV.
Microsoft Store badges using the links you provided.
Starts the Microsoft installation flow directly in the browser/app environment.
Opens the full Microsoft Store product listing in a new browser tab.
Use the badge directly on another page or README.
Designed as a local training and detection application rather than a cloud-only annotation service.
Large-image detection, video detection and Store-style interface screenshots.
The exact badge links are included in the page and can also be copied separately.
<a href="https://get.microsoft.com/installer/download/9pkc4sbbbdj8?referrer=appbadge" target="_self">...</a><a href="https://apps.microsoft.com/detail/9pkc4sbbbdj8?referrer=appbadge&mode=full" target="_blank" rel="noopener noreferrer">...</a>