Release 2025/4/10
Last updated
Last updated
We’re excited to announce new features and improvements for April! These updates enhance annotation workflows, model evaluation, and platform flexibility for all users.
You can now compute VQA metrics across models for boolean, option, and number question types. Gain deeper insight into performance with standardized evaluations.
On VQA Data Slice > Click Run Metrics
Several enhancements to streamline your DataVerse experience:
Data Visualization UI Updates: New preview modes and smoother page navigation.
VQA Filtering Options: Filter questions based on answer type or whether they have been answered.
Annotation Labeling Panel Enhancements:
Clearer visual styles and layouts
More information on images and reviewers
Annotation List View Adjustments: Improved display content for better readability.
While annotating, you can now adjust auto-generated BBox labels for target classes with a single click—making annotation more accurate and flexible.
Support for point and polyline annotations is now available, expanding your ability to label various data types.
You can now send datasets or images across different projects, offering greater flexibility in data management and collaboration.
The SDK has been upgraded to support data import/export workflows more efficiently, helping teams automate and scale operations with ease.
VQA Prediction:
We’re working on VQA prediction capabilities, allowing you to run model inference directly within your project—turning annotated data into insights faster than ever.
Be sure to follow us for more details and updates as we approach the release date.
New hotkeys for faster annotation
For technical support, please contact our during business hours.
