8.1.1.6.1.3. blueoil.metrics.object_keypoint_similarity

8.1.1.6.1.3.1. Module Contents

8.1.1.6.1.3.1.1. Functions

compute_object_keypoint_similarity(joints_gt, joints_pred, image_size=(160, 160))

Compute a object keypoint similarity for a batch of examples.

_compute_oks(joints_gt, joints_pred, image_size=(160, 160))

Compute a object keypoint similarity for one example.

blueoil.metrics.object_keypoint_similarity.kpt_oks_sigmas
blueoil.metrics.object_keypoint_similarity.variances
blueoil.metrics.object_keypoint_similarity.compute_object_keypoint_similarity(joints_gt, joints_pred, image_size=(160, 160))

Compute a object keypoint similarity for a batch of examples.

Parameters
  • joints_gt – a numpy array of shape (batch_size, num_joints, 3).

  • joints_pred – a numpy array of shape (batch_size, num_joints, 3).

  • image_size – a tuple, (height, width).

Returns

float.

Return type

oks_batch

blueoil.metrics.object_keypoint_similarity._compute_oks(joints_gt, joints_pred, image_size=(160, 160))

Compute a object keypoint similarity for one example.

Parameters
  • joints_gt – a numpy array of shape (num_joints, 3).

  • joints_pred – a numpy array of shape (num_joints, 3).

  • image_size – a tuple, (height, width).

Returns

float.

Return type

oks