SinglePoseDetectionModel

class SinglePoseDetectionModel(internalModel: OnnxInferenceModel, modelKindDescription: String?) : SinglePoseDetectionModelBase<Bitmap> , InferenceModel, CameraXCompatibleModel

SinglePoseDetectionModel is an ultra-fast and accurate model that detects 17 keypoints and 18 basic edges of a body.

It internally uses ONNXModels.PoseDetection.MoveNetSinglePoseLighting or ONNXModels.PoseDetection.MoveNetSinglePoseThunder under the hood to make predictions.

Parameters

internalModel

model used to make predictions

Constructors

SinglePoseDetectionModel

android

fun SinglePoseDetectionModel(modelBytes: ByteArray)

Constructs the pose detection model from a model bytes.

SinglePoseDetectionModel

android

fun SinglePoseDetectionModel(internalModel: OnnxInferenceModel, modelKindDescription: String? = null)

Functions

android

open override fun close()

convert

android

open override fun convert(output: OrtSession.Result): DetectedPose

Converts raw model output to the result.

copy

android

open override fun copy(copiedModelName: String?, saveOptimizerState: Boolean, copyWeights: Boolean): InferenceModel

detectPose

android

fun detectPose(image: Bitmap): DetectedPose

Detects a pose for the given image.

initializeWith

android

open override fun initializeWith(vararg executionProviders: ExecutionProvider)

Initialize the model with the specified executions providers.

predict

android

open override fun predict(inputData: FloatArray): Int

open fun predict(input: Bitmap): DetectedPose

Makes prediction on the given input.

predictSoftly

android

open override fun predictSoftly(inputData: FloatArray, predictionTensorName: String): FloatArray

reshape

android

open override fun reshape(vararg dims: Long)

summary

android

open override fun summary(): ModelSummary

Properties

inputDimensions

android

open override val inputDimensions: LongArray

internalModel

android

open override val internalModel: OnnxInferenceModel

modelKindDescription

android

open override val modelKindDescription: String?

High-level description of the model. Used for model summary printing. For the model hub models it equals to the string representation of the OnnxModelType

preprocessing

android

open override val preprocessing: Operation<Bitmap, Pair<FloatArray, TensorShape>>

Preprocessing operation specific to this model.

targetRotation

android

open override var targetRotation: Int = 0

Target image rotation.