sequenceDiagramparticipant U as User
participant G as GUI
participant R as Radar
participant A as Audio
participant V as Vision
participant F as Fusion
U->>G: Load scenario
G->>R: Build radar runtime
G->>A: Precompute audio windows
G->>V: Load YOLO weights
loop video timeline
G->>V: Run detection (stride) G->>A: Query prediction at t
G->>R: Radar detected state
G->>F: Aggregate modalities
F-->>G: Clear/Watch/Confirmed
end