Analyze this security camera frame for object detection and tracking.

DETECT: {objects}

For EACH detected object, report:
- Type (person/vehicle/animal/etc)
- Position in frame (left/center/right, top/middle/bottom)
- Size estimate (small/medium/large relative to frame)
- Action/state (standing/walking/moving/parked)
- Direction of movement if moving (left/right/toward/away)

RESPONSE FORMAT:
OBJECT_1: type=person, pos=center-bottom, size=medium, state=walking, direction=right
OBJECT_2: type=vehicle, pos=left-top, size=large, state=parked, direction=none

If no {objects} detected, say "NO OBJECTS DETECTED"
Count the total number of each object type.
