VQA Labeling Panel
Last updated
Last updated
The VQA Labeling Panel streamlines annotation by providing tools to answer ontology-defined questions, adjust image settings, toggle grids for spatial precision, and save or submit annotations efficiently, enhancing accuracy and user productivity.
When setting up a VQA Annotation Task, you can configure the batch submission size in the Advanced Settings. This allows users to determine how many annotations can be submitted at once for efficient workflow management.
The VQA Labeling Panel provides a user-friendly interface for annotating visual questions and answers based on the defined ontology. Below are the key features explained:
Answering Questions:
Each question is displayed in accordance with the ontology settings.
Supported answer types include:
Boolean (e.g., True/False)
Number (e.g., Count of objects)
Text (e.g., Descriptive answers)
Option (Dropdown) (e.g., Vehicle types)
Ensure that each question is answered before saving or submitting.
Deplicate:
The Grid feature allows users to toggle a grid overlay on the image to assist with object size estimation and spatial alignment. (Each grid cell is set to 5px x 5px)
Grid on/off:
Zoom, Brightness, and Contrast Adjustment:
Use the image toolbar to:
Zoom in or out of the image.
Keyboard Shortcuts:
Leave Comments:
Provide feedback or remarks for specific annotations using the Comment section.
Save and Submit:
Use Save Annotation to temporarily save progress.
Use Submit to finalize and send the annotations for review or storage.
Switch to Grid Display to view .
Adjust brightness and contrast for better visibility of image details.
Quickly navigate and interact with the panel using shortcuts: