Weapons

    CCTV handgun detection: a labelled night-time clip

    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.

    Handgun in the right hand, tracked by a weapon detector box with a confidence score of 0.3 to 0.4.

    What the clip shows

    The camera is fixed high on the corner of a shop and looks down a lit pavement at night. Parked cars line the kerb and a street runs beyond them. This is a typical overhead CCTV view: a long, narrow field of view, a person who is small in the frame, and artificial light from the storefront.

    A person in a dark jacket, a cap and grey trousers stands for a moment and then walks along the storefront, away from the camera. In this version of the clip a handgun is in the person's right hand for the whole six seconds. The weapon is small compared with the person, and the person is seen from behind as they walk away.

    A red detector box labelled weapon follows the handgun from frame to frame. Each box carries a confidence score, and in this clip the scores sit between 0.3 and 0.4. They are low, which shows what a detector reports when it is only partly sure. The overlay is the output of a trained detector, drawn onto the video for illustration. It is not the label format that ships with the dataset.

    The original footage, with no weapon, is shown lower on this page so you can compare the two. The handgun was generated into the empty scene, and the camera view, the lighting and the person's movement are the same in both clips.

    Camera
    Fixed, mounted high on a shop corner, looking down the pavement
    Scene
    Lit pavement at night, parked cars and a street beyond
    Event
    Person walks away from the camera with a handgun in the right hand
    Classes
    person, handgun, rifle, shotgun, sniper_rifle (dataset)
    Clip length
    6 seconds, 24 fps
    Resolution
    464 x 688

    The original footage

    The generated clip above uses this same camera and scene.

    Dataset

    The dataset this clip belongs to

    Sample frame from the CCTV Weapon Detection datasetWeapons & threats

    CCTV Weapon Detection

    Train AI weapon detection on 1,000 synthetic CCTV images: handguns, rifles, shotguns and look-alike negatives, labelled in YOLO and COCO. Free sample included.

    Size
    1,000 images
    Format
    YOLO + COCO
    View dataset

    Need this event on your own cameras?

    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.