Reviewing recorded CCTV footage
The SCVD demonstration analyses an uploaded video, classifies temporal segments and prepares a log of candidate events and timestamps. It supports human review; the current model does not ingest live streams or identify an aggressor or establish personal responsibility.

