Tutorial on How to Ask Queries
The VTT queries can be image-based or scene-based in a storyline. The system integrates information from multiple cameras over time to answer queries about the scene and events.
Define Objects of Interest
Tab: Add Objects
- - Specify an object of interest by drawing a bounding box or a point.
- - Give it a name so that you can ask queries about it later.
In this example, we draw an bounding box around a person and name it John.
A query consists one or more attribute and/or relationship predicates.
Don't know how to ask? No worries, our tool provides you with suggestions generated from the ontology.
Select an defined object and choose an attribute from the list.
Select two defined objects and choose a relationship to ask from the list.
If you make multiple selections, predicates will be conjuncted automatically.
Tab: Query > Ontology
The ontology graph on the right will highlight the paths you have explored.
Tab: Query > Storyline
The storyline tab on the right will display the system's answers.
Natural Language Queries
You can also type here to interact with the system asking natural language questions.
Parsing results generated by our computer vision modules.
Define objects of interest before asking queries about them.
Compose queries and see how the system answers them.