robokudo.types.human

Human-related annotation types for Robokudo.

This module provides types for representing various aspects of human detection and analysis in computer vision applications. It includes:

  • Face detection results

  • Facial expressions and mimics

  • Human attributes

  • Activity recognition

  • Body keypoint detection

The types support both 2D and 3D representations where applicable.

Classes

FaceAnnotation

Face detection result combining 2D, 3D and pose information.

MimicAnnotation

Facial expression or mimic detection result.

AttributeAnnotation

Generic human attribute annotation.

ActivityAnnotation

Human activity detection result.

KeypointAnnotation

Human body keypoint detection result.

Module Contents

class robokudo.types.human.FaceAnnotation

Bases: robokudo.types.cv.ImageROI, robokudo.types.cv.Points3D, robokudo.types.tf.Pose

Face detection result combining 2D, 3D and pose information.

Provides: * 2D region of interest in image * 3D point cloud of face * 6-DOF pose of face

Note

Does not contain identity information. Use Classification annotations on the parent HumanAnnotation for identity.

Variables:

  • mask – Binary mask of face region

  • roi – Rectangular face region

  • points – 3D point cloud of face

  • rotation – Face orientation as quaternion [x,y,z,w]

  • translation – Face position [x,y,z]

class robokudo.types.human.MimicAnnotation

Bases: robokudo.types.core.Annotation

Facial expression or mimic detection result.

Represents detected facial expressions or mimics.

Variables:

type – Type of expression/mimic detected

type = ''
class robokudo.types.human.AttributeAnnotation

Bases: robokudo.types.core.Annotation

Generic human attribute annotation.

Used for various human attributes such as: * Age * Height * Clothing * Gender * Other physical or visual characteristics

Variables:

type – Type of attribute

type = ''
class robokudo.types.human.ActivityAnnotation

Bases: robokudo.types.core.Annotation

Human activity detection result.

Represents detected human activities and their context. Can be subclassed for specific activity types.

Examples: * Pointing * Waving * Walking * Sitting

Variables:

  • type – Type of activity detected

  • interaction_with – Target of interaction if applicable

type = ''
interaction_with = None
class robokudo.types.human.KeypointAnnotation

Bases: robokudo.types.core.Annotation

Human body keypoint detection result.

Represents detected keypoints like: * Joint positions * Body part locations * Facial landmarks

Supports both 2D and 3D keypoint types.

Variables:

  • KP_TYPE_2D – Identifier for 2D keypoint sets

  • KP_TYPE_3D – Identifier for 3D keypoint sets

  • keypoints – List of detected keypoints

  • type – Type of keypoints (2D or 3D)

KP_TYPE_2D = '2D'
KP_TYPE_3D = '3D'
keypoints = []
type = None