What is this?


Tracking objects using target bounding box.

Subscribing Topic

  • ~input (sensor_msgs/Image)

    Raw image.

  • ~input/rects (jsk_recognition_msgs/RectArray)

    Rectangles on an input image.

  • ~input/class (jsk_recognition_msgs/ClassificationResult)

    Class labels for each rectangles.

Publishing Topic

  • ~output/labels (jsk_recognition_msgs/LabelArray)

    Tracking id labels of each tracked objects.

  • ~output/viz (sensor_msgs/Image)

    Visualized image of recognition result.


  • ~gpu (Int, default: 0)

    GPU id.

  • ~target_labels (List of String, default: None)

    If this param is specified, tracking only specified label class. If this value is None, use all label.

  • ~pretrained_model (String, required)

    Pretrained model path.

  • ~approximate_sync (Bool, default: False)

    Whether to use approximate for input topics.

  • ~queue_size (Int, default: 100)

    How many messages you allow about the subscriber to keep in the queue. This should be big when there is much difference about delay between two topics.

  • ~slop (Float, default: 0.1)

    How many seconds you allow about the difference of timestamp when you specify ~approximate_sync.


roslaunch jsk_perception sample_deep_sort_tracker.launch gpu:=0