Behaviour and safety
In a hospital lift, a nurse clutches their chest, slides down the wall and ends sitting collapsed on the floor.
The camera is a fisheye on the ceiling of a hospital lift. It sees the whole small space at once, from directly above, with the curved edges typical of the lens. A patient lies on a gurney in the lift, and a nurse in blue scrubs, a cap and a mask stands beside it.
The event is a collapse. The nurse clutches their chest, slides down against the wall of the lift and ends sitting collapsed on the floor. It happens within the six seconds of the clip and ends with the person sitting against the wall, not lying flat, so a system that only looks for a person lying down would be late.
A lift is a good test for fall detection because the space is tight, the lighting is flat, and the frame holds other things besides the person who falls. Here the gurney and the patient are in view and must not be mistaken for the event. From above, a seated person on the floor and a standing person look different mainly in how much of the floor they cover.
The clip is entirely synthetic and no real person appears in it. The dataset behind it has 1,000 images with person and fallen_person classes in YOLO and COCO formats, and a free sample of 113 images is published on Kaggle and Hugging Face.
Dataset
Behaviour & safetySynthetic fall detection dataset: 1,000 CCTV images of fallen people plus sitting and kneeling look-alikes, labelled person / fallen_person in YOLO and COCO.
More clips
A person walks along a lit shop front at night with a handgun in the right hand, followed frame by frame by a red weapon box.
A person turns and walks past a shop door at night with a large kitchen knife held low in the right hand.
From a moving drone, three civilians walk a dirt track and a group of armed figures enters from the left.
Send 5–20 photos of your empty scenes and we render the event into them, as images or video clips with boxes, tracking IDs and event times.