High level api¶

Face detection and attributes estimation¶

High level api is represented VLFaceDetection and VLFaceDetector classes.

VLFaceDetection is container for a face detection. It has properties which are corresponding face estimations. Every property is a singleton which estimates corresponding attribute once and remembers it for next calling.

VLFaceDetector is a wraper on FaceDetector which convert FaceDetection to VLFaceDetection. Class contains a face estimator collection as a class attribute for a detections attributes estimation. If you want specify an unique VLFaceEngine for the instance of VLFaceDetector you should set the FaceEngine in the init of the class.

Example¶

>>> from lunavl.sdk.luna_faces import VLFaceDetector
>>> from lunavl.sdk.luna_faces import VLImage
>>> detector = VLFaceDetector()
>>> image = VLImage.load(
...         url='https://cdn1.savepice.ru/uploads/2019/4/15/aa970957128d9892f297cdfa5b3fda88-full.jpg')
>>> detection = detector.detectOne(image)
>>> detection.basicAttributes.age
16.0
>>> detection.emotions.predominateEmotion.name
'Happiness'

Module realize hight level api for estimate face attributes

class lunavl.sdk.luna_faces.VLFaceDetection(coreDetection, image, estimatorCollection)[source]¶

High level detection object. .. attribute:: estimatorCollection

collection of estimators

type

FaceEstimatorsCollection

_emotions¶

lazy load emotions estimations

Type: Optional[Emotions]

_eyes¶

lazy load eye estimations

Type: Optional[EyesEstimation]

_mouthState¶

lazy load mouth state estimation

Type: Optional[MouthStates]

_basicAttributes¶

lazy load basic attribute estimation

Type: Optional[BasicAttributes]

_gaze¶

lazy load gaze direction estimation

Type: Optional[GazeEstimation]

_warpQuality¶

lazy load warp quality estimation

Type: Optional[Quality]

_headPose¶

lazy load head pose estimation

Type: Optional[HeadPose]

_ags¶

lazy load ags estimation

Type: Optional[float]

_transformedLandmarks68¶

lazy load transformed landmarks68

Type: Optional[Landmarks68]

ags¶

Get ags of the detection.

Returns: emotions
Return type: float

asDict()[source]¶

Convert to dict.

Returns: All estimated attributes will be added to dict
Return type: Dict[str, Union[dict, list, float]]

basicAttributes¶

Get all basic attributes of the detection.

Returns: basic attributes (age, gender, ethnicity)
Return type: BasicAttributes

descriptor¶

Get a face descriptor from warp

Returns: mouth state
Return type: FaceDescriptor

emotions¶

Get emotions of the detection.

Returns: emotions
Return type: Emotions

eyes¶

Get eyes estimation of the detection.

Returns: eyes estimation
Return type: EyesEstimation

gaze¶

Get gaze direction.

Returns: gaze direction
Return type: GazeEstimation

headPose¶

Get a head pose of the detection. Estimation bases on 68 landmarks

Returns: head pose
Return type: HeadPose

mouthState¶

Get a mouth state of the detection

Returns: mouth state
Return type: MouthStates

warp¶

Get warp from detection.

Returns: warp
Return type: Warp

warpQuality¶

Get quality of warped image which corresponding the detection :returns: quality

Return type: Quality

class lunavl.sdk.luna_faces.VLFaceDetector(detectorType=<DetectorType.FACE_DET_DEFAULT: 'Default'>, faceEngine=None)[source]¶

High level face detector. Return VLFaceDetection instead simple FaceDetection.

estimatorCollection¶

face estimator collections for new detections.

Type: FaceEstimatorsCollection

_faceDetector¶

face detector

Type: FaceDetector

faceEngine¶

face engine for detector and estimators, default FACE_ENGINE.

Type: VLFaceEngine

detect(images, limit=5)[source]¶

Batch detect faces on images.

Parameters

images – input images list. Format must be R8G8B8
limit – max number of detections per input image

Returns

return list of lists detection, order of detection lists is corresponding to order input images

Return type

List[List[VLFaceDetection]]

detectOne(image, detectArea=None)[source]¶

Detect just one best detection on the image.