Skip to content

Latest commit

 

History

History
489 lines (306 loc) · 7.35 KB

node-catalog.mdx

File metadata and controls

489 lines (306 loc) · 7.35 KB
title description
Node Catalog
An overview of every node supported by the Kinetix ML platform and supporting libraries.

Key Point

Pose Detection 2D PoseDetection2D

Detects 17 key points on a person's body including their major joints, eyes, ears, and nose. Helpful for analyzing body motion including body poses and athletic movements.

  • Parameters:

    • Confidence Threshold (Double): 0.3

  • Inputs:

    • Image

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Face Mesh Detection FaceMeshDetection

Detects 17 key points on a person's body including their major joints, eyes, ears, and nose. Helpful for analyzing body motion including body poses and athletic movements.

  • Inputs:

    • Image

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Calculate Joint Angles CalcJointAngles

Calculates the 2D angle of each joint.

  • Inputs:

    • KPFrame

  • Outputs:

    • Angles (Vec)
    • Labels (Label[])

    Platforms:JS

Three Key Point Angle ThreeKPAngle

Calculates the 2D angle between three key points.

  • Inputs:

    • KeyPoint
    • KeyPoint
    • KeyPoint

  • Outputs:

    • Angle (Double)

    Platforms:JS

Key Point Distance KPDist

Calculates the distance between two key points.

  • Inputs:

    • KeyPoint
    • KeyPoint

  • Outputs:

    • Dist (Double)

    Platforms:JS

Smooth Key Points SmoothKeyPoints

Performs a moving weighted average smoothing algorithm that prioritizes model confidence and weeds out outliers.

  • Parameters:

    • Frame Count (Int): 10

  • Inputs:

    • KPFrame

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Crop Key Points CropKeyPoints

Performs a square crop around the region of the KPFrame for use with normalization.

  • Inputs:

    • KPFrame

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Normalize Key Points NormKeyPoints

Performs a L2 normalization on a cropped KPFrame.

  • Inputs:

    • KPFrame

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Normalize Key Points Size NormKeyPointsSize

Performs a normalization on a cropped KPFrame based on the width and height of the keypoints.

  • Inputs:

    • KPFrame

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

Compare KPFrames CompareKPFrames

Performs a keypoint-wise cosine similarity on all the keypoints in two KPFrames.

  • Inputs:

    • KPFrame
    • KPFrame

  • Outputs:

    • Similarities (Vec)

    Platforms:JS

Calculate Key Point Velocities CalcKeyPointVelocities

Calculates the velocity of each key point using the last frame as reference point.

  • Inputs:

    • KPFrame

  • Outputs:

    • Velocities (Vec)

    Platforms:JS

Create Key Point CreateKeyPoint

Creates a key point object.

  • Inputs:

    • Double
    • Double
    • Double
    • String

  • Outputs:

    • Key Point (KeyPoint)

    Platforms:JS

Get Key Point GetKeyPoint

Gets a key point object from a KPFrame.

  • Parameters:

    • Index (Int): 0

  • Inputs:

    • KPFrame

  • Outputs:

    • Key Point (KeyPoint)

    Platforms:JS

Deconstruct Key Point DeconstructKeyPoint

Deconstructs a key point object.

  • Inputs:

    • KeyPoint

  • Outputs:

    • x (Double)
    • y (Double)
    • score (Double)
    • name (String)

    Platforms:JS

Set Key Point SetKeyPoint

Sets a key point object.

  • Parameters:

    • Index (Int): 0

  • Inputs:

    • KPFrame
    • KeyPoint

  • Outputs:

    • KP Frame (KPFrame)

    Platforms:JS

CreateCrop CreateCrop

Creates a crop rectangle from a KPFrame.

  • Inputs:

    • KPFrame
    • Image

  • Outputs:

    • crop (Vec)

    Platforms:JS

Object Detection

Roboflow Detect RoboflowDetect

Runs object detection on an input image using Roboflow Object Detection models.

  • Parameters:

    • Model (String):
    • Version (Int): 1
    • API Key (String):
    • Confidence Threshold (Double): 0.5
    • Overlap Threshold (Double): 0.5

  • Inputs:

    • Image

  • Outputs:

    • BBox Frame (BBoxFrame)

    Platforms:JS

Vector

Smooth Vectors SmoothVecs

Performs a moving average smoothing algorithm on vectors.

  • Parameters:

    • Frame Count (Int): 10

  • Inputs:

    • Vec

  • Outputs:

    • KP Frame (Vec)

    Platforms:JS

Get Vector Value GetVecValue

Gets a value from a Vec.

  • Parameters:

    • Index (Int): 0

  • Inputs:

    • Vec

  • Outputs:

    • Double (Double)

    Platforms:JS

Set Vector Value SetVecValue

Sets a value to a Vec.

  • Parameters:

    • Index (Int): 0

  • Inputs:

    • Vec
    • Double

  • Outputs:

    • Vec (Vec)

    Platforms:JS

Display

Draw Key Points DrawKeyPoints

Draws a KPFrame on an input canvas.

  • Parameters:

    • Radius (percent) (Double): 0.1

  • Inputs:

    • KPFrame
    • Image
    • Canvas

    Platforms:JS

Draw Bounding Boxes DrawBBoxFrame

Draws a BBoxFrame on an input canvas.

  • Inputs:

    • BBoxFrame
    • Canvas

    Platforms:JS

Draw Labels DrawLabels

Draws Labels on an input canvas.

  • Inputs:

    • Label[]
    • Image
    • Canvas

    Platforms:JS

Draw One Label DrawLabel

Draws one Label on an input canvas.

  • Inputs:

    • Label
    • Image
    • Canvas

    Platforms:JS

Create Label CreateLabel

Creates a label object.

  • Inputs:

    • Double
    • Double
    • String

  • Outputs:

    • Label (Label)

    Platforms:JS

Math

Add Inputs AddInputs

Adds two inputs.

  • Inputs:

    • Double
    • Double

  • Outputs:

    • Result (Double)

    Platforms:JS

Divide Inputs DivideInputs

Divides two inputs.

  • Inputs:

    • Double
    • Double

  • Outputs:

    • Result (Double)

    Platforms:JS

Multiply Inputs MultiplyInputs

Multiply two inputs.

  • Inputs:

    • Double
    • Double

  • Outputs:

    • Result (Double)

    Platforms:JS

Subtract Inputs SubtractInputs

Subtracts two inputs.

  • Inputs:

    • Double
    • Double

  • Outputs:

    • Result (Double)

    Platforms:JS

Constant Constant

Exports its parameter as an output for use with other nodes.

  • Parameters:

    • Value (String): 0
    • Data Type (String): 0

  • Outputs:

    • Data (Any)

    Platforms:JS

Round Round

Rounds a Double to the closest whole number.

  • Inputs:

    • Double

  • Outputs:

    • rounded (Int)

    Platforms:JS

Clamp Clamp

Clamps a value between a minimum and maximum.

  • Parameters:

    • Minimum (Double): 0
    • Maximum (Double): 100

  • Inputs:

    • Double

  • Outputs:

    • clamped (Int)

    Platforms:JS

Logic

Conditional Conditional

Outputs a boolean based on the comparison of two inputs

  • Parameters:

    • Operator (String): ==

  • Inputs:

    • Any
    • Any

  • Outputs:

    • Boolean (Boolean)

    Platforms:JS

Switch Switch

Prevents execution based on a boolean value. The second input passes directly to the Passthrough output for use in the node after.

  • Inputs:

    • Boolean
    • Any

  • Outputs:

    • Passthrough (Any)

    Platforms:JS

I/O

Crop Crop

Rounds a Double to the closest whole number.

  • Inputs:

    • Vec
    • Image

  • Outputs:

    • cropped (Image)

    Platforms:JS