2023 RELEASES
Unitlab product updates released in 2023.
Unitlab product updates released in 2023.
Features
Updated private storage handling, AI data upload, batch auto-labeling, and synchronized SDK/CLI behavior.
Added more consistent source URLs and annotation result handling across stored data.
Fixes
Improved storage error handling, serializer performance, and malformed or missing source handling.
Removed obsolete SAM decoder paths and corrected batch-processing edge cases.
Features
Introduced workspace management with Owner, Manager, and Member roles, invitations, project membership, and workspace-scoped access.
Moved Unitlab to subscription-based billing with Stripe checkout, annual plans, customer portal support, usage quotas, and labeling-service billing.
Added Segment Anything integration, annotation rotation support, and broader dataset lifecycle controls.
Fixes
Corrected workspace permissions, invitations, statistics, comments, cloning, and Stripe error paths.
Improved long-term platform support and strengthened encrypted-field handling.
Features
Added improved polygon and segmentation rendering, history managers, AI-model request headers, gzip responses, and batch auto-label status updates.
Fixes
Corrected segmentation merging, empty results, thumbnail holes, opacity, and source-field naming.
Increased the default API page size and cleaned up rendering utilities.
Features
Added COCO-style segmentation, web and CLI dataset downloads, presigned URLs, background cloning, and stronger AI-model class handling.
Standardized product terminology around Projects, annotation types, markup types, and source fields across the API, SDK, and CLI.
Fixes
Improved dataset release and clone performance, grouped ZIP creation, segmentation thumbnails, and AI response conversion.
Corrected key errors, temporary-file handling, conversion noise, and social-auth behavior.
Features
Expanded dynamic data-type support, COCO export behavior, dataset cloning, and release dataset data responses.
Added richer AI-model metadata and data-type-aware task behavior.
Fixes
Corrected polygon drawing, scaling, thickness, thumbnails, metadata defaults, ordering, and empty-list responses.
Simplified model constraints and removed duplicate or obsolete thumbnail handling.
Features
Added dataset size, description, time statistics, deletion, COCO export, S3 ZIP delivery, and project-independent release data.
Added AI-model create, update, delete, validation, preview, upload, public/private checks, and monthly usage limits.
Fixes
Corrected dataset and reviewer statistics, class assignment, passwords, release data, and empty-result handling.
Strengthened upload validation and source metadata checks.
Features
Added whole-image and point annotation support, colored bounding boxes, reviewer timing, seen status, and monthly API statistics.
Expanded API-key reporting, task-level image statistics, and thumbnail responses.
Fixes
Corrected annotation conditions, null handling, time-zone awareness, and review-state updates.
Features
Added dataset pages, dataset cloning, secure data URLs, grouped auto-labeling, cloned-history merging, and AI-result merging.
Added dynamic AI-model classes and statistics, Auto Crop, common task classes, API keys, tagged classes, and expanded SDK/CLI endpoints.
Fixes
Corrected segmentation payloads, mask merging, counters, thumbnails, clone ordering, attributes, and key errors.
Hardened paid-task history creation, CORS, deployment timeouts, throttling, and review-status handling.
Features
Added release and dataset lists, downloads, rendered annotation metadata, task types, paid-task checkout, API-key checks, and AI-release progress.
Added labeling access, available classes, task statistics, and Stripe test/live webhook handling.
Fixes
Improved segmentation rendering speed, thumbnail delivery, resize behavior, N+1 queries, release output, and key-error handling.
Explore related Unitlab capabilities: enterprise data annotation workflows · Unitlab’s image annotation platform