Tags

v9.0.0

Frontend:
- Added support for [CVAT](https://www.cvat.ai/) as a labeling tool:
  - When creating a new Dataset, select the new `CVAT` dataset type.
- Added support for `RTMDet INS - X` Model for Instance Segmentation.
- Reworked "Model Testing" Page:
  - Now shows all previous prediction runs in a table.
  - Clicking on a prediction run now shows all predicted images, with the option to open a larger version of the image.
  - Moved the form from the old page to the "Predict with model" button.
- Reworked the image viewer to allow navigating through all images:
  - You can now use the buttons on the side to navigate or use the Left and Right Arrow keys.
- The tables on all main pages (e.g., Data Management) have been updated to automatically update every 30 seconds.
- Improved the loading spinner layout on all main pages to correctly overlay the tables and be centered.
- You can now specify the recording interval for recordings via the edge device or add pictures manually.
- You can now see auto-annotation jobs via the new "Auto Annotations" page.

Backend:
- Add support for [CVAT](https://www.cvat.ai/) as a labeling tool:
  - Refactored the LabelingService class to work with multiple Labeling Services: CVAT and LabelStudio.
  - Introduce a new dataset postgres format
  - Added a dedicated endpoint for migrating datasets.
- Add `mlcvzoo-rf-detr` to support training and inference for Object Detection and Instance Segmentation of rfdetr models
- Refactor Model Testing
  - Load information about model testing from dedicated mlflow runs (prediction jobs)
  - Provide this information via dedicated endpoints
- Add the possibility to load information about Auto Annotation jobs