Annotaforge: Dataset Lab

Annotaforge is a fully offline image-box annotation desk with user-defined class selection, drag-to-label geometry, undo, normalized coordinates, review counts, and JSON export. WHAT IT DOES - Choose a local image without uploading it - Draw multiple rectangular annotations with one of five visible classes - Undo annotations and review box and class counts - Export normalized image-space geometry as JSON PRIVATE BY DEFAULT No account, advertising, tracking, analytics SDK, or subscription. Workspace data stays in the app container. Imports, scans, device access, and network actions happen only when you request them. ONE PURCHASE The complete core tool, local workspace, records, analysis, export, and offline reference are included in the paid download. IMPORTANT BOUNDARY Annotations and QA findings require human review. The app does not train or evaluate a model.