vpi.KLTFeatureTracker

class vpi.KLTFeatureTracker(img_template: vpi.Image, in_boxes: vpi.Array, *, backend: vpi.Backend = vpi.Backend.DEFAULT) → vpi.KLTFeatureTracker
class vpi.KLTFeatureTracker(img_template: vpi.Image, in_boxes: vpi.Array, *, max_template_count: int, max_template_size: Tuple[int, int], backend: vpi.Backend = vpi.Backend.DEFAULT) → vpi.KLTFeatureTracker

Creates the basis for the KLT algorithm.

Note

This class allocates all resources needed by the Kanade-Lucas-Tomasi (KLT) feature tracker algorithm. An object of this class is able to run the KLT algorithm via its call operator.

Parameters

  • img_template (vpi.Image) – Template image.

  • in_boxes (vpi.Array) – Input bounding boxes.

  • max_template_count (int, optional) – Maximum number of templates to be tracked.

  • max_template_size (Tuple[int, int], optional) – Maximum width/height of each tracked template.

  • backend (vpi.Backend, optional) – The backend to be used by the algorithm.

Returns

The main object of the KLT algorihtm.

Return type

vpi.KLTFeatureTracker

Caution

Restrictions to several arguments may apply. Check the C API references of the create payload function and the group concepts for more details.

Methods

__call__(img_reference, *[, ...])

Runs a KLT Feature tracker on two images.

add_boxes(self, bboxes)

Add input bounding boxes to be tracked.

default_update(in_boxes, in_preds, ...)

Get the default update function.

in_predictions(self)

Get current input predictions.

out_estimations(self)

Get current output estimations.