API

The located Function under com.tenginekit.

init

Parameter

context: context in activity，
config: AndroidConfig is the config of work

    KitCore.init(Context context, AndroidConfig config);

For example, set the default configuration like this.

    KitCore.init(context,
            AndroidConfig
                .create()
                .setCameraMode()
                .setDefaultFunc()
                .setDefaultInputImageFormat()
                .setInputImageSize(previewWidth, previewHeight)
                .setOutputImageSize(OutputWidth, OutputHeight)
        );

AndoridConfig

.create() First create an AndroidConfig.

Configuration mode:

.setNormalMode() Set to normal mode
.setCameraMode() Set to camera mode(This mode will automatically rotate the picture according to the front and rear camera)
.setHandleMode(HandleMode mHandleMode) Parameter transmission mode setting mode

    public enum HandleMode{
        Normal, // Normal mode
        Camera // Camera mode
    }

Configure input picture format:

.setDefaultInputImageFormat() Set the default picture format (YUV_NV21)
.setInputImageFormat(ImageFormat imageFormat) Parameter transmission mode to set input picture format

    public enum ImageFormat
    {
        YUV_NV21,
        RGB,
        RGBA,
        GRAY
    }

Configure input and output:

.setInputImageSize(int w, int h) Input width and height (preview_w, preview_h)
.setOutputImageSize() Output width and height (display size)

Configuration function (functions are based on image detection):

.setDefaultFunc() Set default functions (face detection and key points)
.openFunc(Func func) Pass parameter setting function (image detection must be set to AndroidConfig.Func.Detect or AndroidConfig.Func.BlazeFace)

    public enum Func
    {
        Detect, // 
        Landmark,
        Attribution,
        BlazeFace,
        FaceMesh,
        Iris,
        BlazePose,
        BlazePoseLandmark,
        HandDetect,
        HandLandmark3d,
    }

enum	Function
Detect	Face detect
Landmark	Face landmarks(212)
Attribution	Face attribution
BlazeFace	Face detect
FaceMesh	Face 3d landmarks(468)
Iris	Iris landmarks
BlazePose	Body detect
BlazePoseLandmark	Body landmarks
HandDetect	Hand detect
HandLandmark3d	Hand landmarks

Tips：BlazeFace, FaceMesh, Iris, BlazePose, BlazePoseLandmark, HandDetect, HandLandmark3d, models are from Google，The project address is：https://github.com/google/mediapipe

get face detect infos

Since all functions on face are based on face detection, first create an object of Face.FaceDetect. Detect detection will be faster, BlazeFace will be more accurate and the angle can be larger (BlazeFace is based on the Google model). Will eventually return a FaceDetectInfo list;

Parameter

imageData: Input data

    Face.FaceDetect faceDetect = Face.detect(imageData);
    List<FaceDetectInfo> faceDetectInfos = faceDetect.getDetectInfos();

get landmark infos(212 landmarks)

The face key point function is based on face detection, so the landmark information acquisition method should be based on the Face.FaceDetect object created earlier. Finally returns a FaceLandmarkInfo list;

    List<LandmarkInfo> landmarkInfos = faceDetect.landmark2d();

get 3D landmark infos(This feature is based on Google model)(468 landmarks)

The face key point function is based on face detection, so the 3d landmark information acquisition method should be based on the Face.FaceDetect object created earlier. Finally returns a FaceLandmark3dInfo list;

    List<FaceLandmark3dInfo> landmarkInfos = faceDetect.landmark3d();

get attribution infos

The attribute function is based on face detection, so the method of obtaining attribute information is based on the Face.FaceDetect object created earlier. Finally returns a FaceAttributionInfo list;

    List<AttributionInfo> attributionInfos = faceDetect.attribution();

get iris information (this function is based on Google's model) (76 key points)

The iris function is based on face detection and face 3d key points, so iris3d information acquisition should be based on the Face.FaceDetect object created earlier. Finally return a FaceIrisInfo list; 76 key points include 5 iris key points and 71 key points around the eyes.

    List<FaceIrisInfo> irisInfos = faceDetect.iris3d();;

get hand detection information(this function is based on Google's model)

Since all the functions of the hand are based on hand detection, first create an object of Hand.HandDetect. Will eventually return a List;

Parameters

imageData: input data

    Hand.HandDetect handDetect = Hand.detect(imageData);
    List<HandDetectInfo> handDetectInfos = handDetect.getDetectInfos();

get hand landamrk info (this function is based on Google's model)(21 key points)

The hand key point function is based on hand detection, so the landmark information acquisition method should be based on the Hand.HandDetect object created earlier. Finally returns a HandLandmarkInfo;

     List<HandLandmarkInfo> landmarkInfos = handDetect.landmark3d();

get body detection information(this function is based on Google's model)

Since all the functions of the body are based on body detection, first create an object of Body.BodyDetect. Will eventually return a List;

Parameters

imageData: input data

    Body.BodyDetect bodyDetect = Body.detect(imageData);
    List<BodyDetectInfo> bodyDetectInfos = bodyDetect.getDetectInfos();

get body landamrk info (this function is based on Google's model)(25 key points)

The body key point function is based on body detection, so the landmark information acquisition method should be based on the Body.BodyDetect object created earlier. Finally returns a BodyLandmarkInfo;

     List<BodyLandmarkInfo> landmarkInfos = bodyDetect.landmark2d();

switch camera

Switching cameras is only useful when in camera mode.

Parameter

back:Whether it is a rear camera

    Face.Camera.switchCamera(boolean back);

set rotation

Setting rotation is only useful in camera mode.

Parameter

ori: Rotation angle (vertical to 0°)
isScreenRotate: whether the display area follows rotation
outputW: show width
outputH: show height

    Face.Camera.setRotation(int ori, boolean isScreenRotate, int outputW, int outputH);

release

    FaceManager.getInstance().release();

Data structure

FaceDetectInfo

Parameter name	Parameter type	Comment
top	int	Distance display upper edge distance
bottom	int	Distance display bottom edge distance
left	int	Distance shows the distance of the left edge
right	int	Distance display right edge distance
width	int	Face rect width
height	int	Face rect height

.asRect() is a method in FaceDetectInfo, which can directly return an object of class Rect.

FaceLandmarkInfo

Parameter name	Parameter type	Comment
landmarks	List	Face key point information 212 points
pitch	float	Human face pitch direction corner
yaw	float	Human face yaw direction corner
roll	float	Human face roll direction corner
leftEyeClose	float	Left eye closure confidence 0~1
rightEyeClose	float	Right eye closure confidence 0~1
mouseClose	float	Mouth closure confidence 0~1
mouseOpenBig	float	Open mouth Big confidence 0~1

FaceLandmark3dInfo

Parameter name	Parameter type	Comment
landmarks	List	Face key point information 468 points

FaceAttributionInfo

Parameter name	Parameter type	Comment
age	int	Age
isMan	boolean	Is it male
smile	int	Smile degree0~100
glasses	boolean	Whether to wear glasses
beatyOfManLook	int	Face value from a male perspective
beautyOfWomanLook	int	Face value from a female perspective

FaceIrisInfo

Parameter name	Parameter type	Comment
leftEye	eyeInfo	left eye info
rightEye	eyeInfo	right eye info

eyeInfo

Parameter name	Parameter type	Comment
eyeLandmark	List	71 key points of the eye
eyeIris	List	5 key points of the iris

HandDetectInfo

Parameter name	Parameter type	Comment
top	int	Distance display upper edge distance
bottom	int	Distance display bottom edge distance
left	int	Distance shows the distance of the left edge
right	int	Distance display right edge distance
width	int	Hand rect width
height	int	Hand rect height

HandLandmarkInfo

Parameter name	Parameter type	Comment
landmarks	List	Hand key point information 21 points

BodyDetectInfo

Parameter name	Parameter type	Comment
top	int	Distance display upper edge distance
bottom	int	Distance display bottom edge distance
left	int	Distance shows the distance of the left edge
right	int	Distance display right edge distance
width	int	Body rect width
height	int	Body rect height

BodyLandmarkInfo

Parameter name	Parameter type	Comment
landmarks	List	Body key point information 25 points

TenginekitPoint

Parameter name	Parameter type	Comment
x	float	Point x position
y	float	Point y position

ImageHandleApi

Parameter

bitmap: Android picture
inputX1: input the upper left corner of the picture x
inputY1: input image in the upper left corner y
inputX2: Enter the bottom right corner of the picture x
inputY2: Enter the lower right corner of the picture y
ouputW: width of output picture
outputH: the height of the output picture
rotation: rotation angle
mirror: whether to mirror

    synchronized static public Bitmap convertImage(Bitmap bitmap,
                                    int inputX1, int inputY1, int inputX2, int inputY2,
                                    int outputW, int outputH, int rotation, boolean mirror);

Parameter

data: input picture byte data
imageOutputFormat: output image format
ouputW: width of output picture
outputH: the height of the output picture
rotation: rotation angle
mirror: whether to mirror

    synchronized static public byte[] convertCameraYUVData(byte[] data,
                                    AndroidConfig.ImageFormat imageOutputFormat,
                                    int inputW, int inputH,
                                    int outputW, int outputH, int rotation, boolean mirror

Parameter

data: input picture byte data
inputW: input picture width
inputH: input image height
ouputW: width of output picture
outputH: the height of the output picture
rotation: rotation angle
mirror: whether to mirror

    synchronized static public Bitmap convertCameraYUVData(byte[] data,
                                    int inputW, int inputH,
                                    int outputW, int outputH,
                                    int rotation, boolean mirror

Parameter

data: input picture byte data
imageInputFormat: input image format
imageOutputFormat: output image format
inputW: input picture width
inputH: input image height
inputX1: input the upper left corner of the picture x
inputY1: input image in the upper left corner y
inputX2: Enter the bottom right corner of the picture x
inputY2: Enter the lower right corner of the picture y
ouputW: width of output picture
outputH: the height of the output picture
rotation: rotation angle
mirror: whether to mirror

    synchronized static public byte[] convertImage(byte[] data,
                                    AndroidConfig.ImageFormat imageInputFormat, AndroidConfig.ImageFormat imageOutputFormat,
                                    int inputW, int inputH,
                                    int inputX1, int inputY1, int inputX2, int inputY2,
                                    int outputW, int outputH, int rotation, boolean mirror

Files

Api.md

Latest commit

History

Api.md

File metadata and controls

API

init

Parameter

AndoridConfig

get face detect infos

Parameter

get landmark infos(212 landmarks)

get 3D landmark infos(This feature is based on Google model)(468 landmarks)

get attribution infos

get iris information (this function is based on Google's model) (76 key points)

get hand detection information(this function is based on Google's model)

Parameters

get hand landamrk info (this function is based on Google's model)(21 key points)

get body detection information(this function is based on Google's model)

Parameters

get body landamrk info (this function is based on Google's model)(25 key points)

switch camera

Parameter

set rotation

Parameter

release

Data structure

FaceDetectInfo

FaceLandmarkInfo

FaceLandmark3dInfo

FaceAttributionInfo

FaceIrisInfo

eyeInfo

HandDetectInfo

HandLandmarkInfo

BodyDetectInfo

BodyLandmarkInfo

TenginekitPoint

ImageHandleApi

Parameter

Parameter

Parameter

Parameter