Multimodal AI
Object tracking
Following the same objects as they move through a video.
Example
A sports system follows a ball as it moves across the field.
Why people use it
It connects sightings across time instead of treating every video frame separately.
What you'll hear
“Keep following that ball through the clip.”
What this means for you
Check how the system handles lost tracks and mistaken identity switches.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can tracking lose an object?
- Yes. Blocked views, fast movement and similar-looking objects can break a track.
- Can it confuse two similar objects?
- Yes. Similar appearance or crossing paths can cause it to switch identities.
- What happens when an object leaves the picture?
- The track may end. Recognizing it again later can require an additional matching step.