KeypointDetectionAnnotation

Represents keypoint detection for a single instance: an annotation id, keypoint coordinates, optional per-keypoint scores, and optional feature descriptors. This datatype is used for pose and keypoint detection inputs and outputs within the Telekinesis ecosystem.

Field

Required

Field	Type	Description
id	int	Annotation identifier (int).

Optional

Field	Type	Description
scores	ArrayLike	Per-keypoint response strengths; 1D array, dtype float32.
keypoints	ArrayLike	Keypoint coordinates; shape (N, 2) or flat length 2N, dtype float32.
num_keypoints	int	Number of keypoints (int); inferred from keypoints if omitted.
descriptors	ArrayLike	Flattened feature descriptors (array).

Methods

Method	Description
to_dict()	Returns a dictionary with keys: id (int), and optionally keypoints, scores, num_keypoints, descriptors (as stored).

Example

import numpy as np
from datatypes import datatypes

keypoints = np.array([[10.0, 20.0], [30.0, 40.0], [50.0, 60.0]], dtype=np.float32)
scores = np.array([0.9, 0.8, 0.7], dtype=np.float32)
ann = datatypes.KeypointDetectionAnnotation(
    id=1,
    keypoints=keypoints,
    scores=scores,
    num_keypoints=3,
)

# Accessing keypoint data for pose or visualization.
# to_dict() returns id and optionally keypoints, scores, num_keypoints, descriptors; or use attributes directly.
d = ann.to_dict()
print("KeypointDetectionAnnotation id:", d["id"])
print("KeypointDetectionAnnotation keypoints shape:", getattr(ann.keypoints, "shape", None), "(N, 2)")
print("KeypointDetectionAnnotation scores shape:", getattr(ann.scores, "shape", None))
print("KeypointDetectionAnnotation num_keypoints:", ann.num_keypoints)