# Interface: Hand

Hand detection

## Properties

### debug

• `Optional`

**debug**: `Object`

#### Type declaration

Name | Type |
---|---|

`anchors` | `Coord2D` [] |

`box` | `PalmBox` |

`circle` | `Coord2D` [] |

`tensors?` | { `edge` : `Uint8Array` ; `mask` : `Uint8Array` ; `rgb` : `Uint8Array` } |

`tensors.edge` | `Uint8Array` |

`tensors.mask` | `Uint8Array` |

`tensors.rgb` | `Uint8Array` |

### handedness

• **handedness**: `number`

Classification score of handedness

Number between -1 and 1 that represents handedness. Negative value signals right hand, positive - left. Bigger magnitude (absolute value) means more rebust classification, the closer value is to zero - the more ambiguous is distinction of the handedness.

### points

• **points**: `HandPoint`

[]

List of hand keypoints

2D pixel coordinate - point in the screen coordinate space. XY coordinates are normalized screen coordinates (scaled by image width and height), while the Z coordinate is depth in orthographic projection space, it has the same scale as X. 3D metric coordinate - point within 3D space of perspective camera located at the space origin and pointed in the negative direction of the Z-axis. 3D & 2D points are perfectly aligned.

### score

• **score**: `number`

Reliability score, number between 0 and 1

### wrist

• **wrist**: `WristDetection`

Wrist detection

Wrist detection provides 3 lines in screen coordinate space. Middle line defines 2D wrist center and unit direction vector of the wrist. Two wrist edges are defined by 2D points at the end of the wrist along transversal section through the center.