Upgrade to OPENRNDR 0.3.47, Javacpp 1.5.5, Tensorflow 0.3.1, Gradle 7.0

build.gradle

@@ -10,7 +10,7 @@ buildscript {
     }
 }
 plugins {
-    id 'org.jetbrains.kotlin.jvm' version '1.4.31'
+    id 'org.jetbrains.kotlin.jvm' version '1.4.32'
 }
 
 def openrndrUseSnapshot = false
@@ -18,15 +18,15 @@ def openrndrUseSnapshot = false
 apply plugin: 'org.jetbrains.dokka'
 
 project.ext {
-    openrndrVersion = openrndrUseSnapshot? "0.4.0-SNAPSHOT" : "0.3.47-rc.2"
+    openrndrVersion = openrndrUseSnapshot? "0.4.0-SNAPSHOT" : "0.3.47"
     kotlinVersion = "1.4.31"
     spekVersion = "2.0.15"
-    libfreenectVersion = "0.5.7-1.5.4"
+    libfreenectVersion = "0.5.7-1.5.5"
-    librealsense2Version = "2.29.0-1.5.4"
+    librealsense2Version = "2.40.0-1.5.5"
     gsonVersion = "2.8.6"
     antlrVersion = "4.9.1"
-    tensorflowVersion = "0.2.0"
+    tensorflowVersion = "0.3.1"
-    mklDnnVersion = "0.21.5-1.5.4"
+    mklDnnVersion = "0.21.5-1.5.5"
 }
 
 switch (org.gradle.internal.os.OperatingSystem.current()) {

gradle/wrapper/gradle-wrapper.properties

@@ -1,5 +1,5 @@
 distributionBase=GRADLE_USER_HOME
 distributionPath=wrapper/dists
-distributionUrl=https\://services.gradle.org/distributions/gradle-6.8.3-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-7.0-bin.zip
 zipStoreBase=GRADLE_USER_HOME
 zipStorePath=wrapper/dists

orx-tensorflow/src/main/kotlin/Tensor.kt

@@ -6,35 +6,38 @@ import org.openrndr.draw.ColorFormat
 import org.openrndr.draw.ColorType
 import org.openrndr.draw.colorBuffer
 import org.openrndr.extra.tensorflow.arrays.*
+import org.tensorflow.Output
 import org.tensorflow.Tensor
 import org.tensorflow.ndarray.StdArrays
 import org.tensorflow.ndarray.buffer.DataBuffers
+import org.tensorflow.op.math.Add
 import org.tensorflow.types.*
 import org.tensorflow.types.family.TType
 import java.nio.ByteBuffer
 import java.nio.ByteOrder
 
-fun ColorBuffer.copyTo(tensor: Tensor<TFloat32>) {
+fun ColorBuffer.copyTo(tensor: TFloat32) {
     val buffer = ByteBuffer.allocateDirect(effectiveWidth * effectiveHeight * format.componentCount * 4)
     buffer.order(ByteOrder.nativeOrder())
     this.read(buffer, targetType = ColorType.FLOAT32)
     buffer.rewind()
     val dataBuffer = DataBuffers.of(buffer.asFloatBuffer())
-    tensor.data().write(dataBuffer)
+    tensor.write(dataBuffer)
 }
 
 @JvmName("copyToTUint8")
-fun ColorBuffer.copyTo(tensor: Tensor<TUint8>) {
+fun ColorBuffer.copyTo(tensor: TUint8) {
     val buffer = ByteBuffer.allocateDirect(effectiveWidth * effectiveHeight * format.componentCount)
     buffer.order(ByteOrder.nativeOrder())
     this.read(buffer, targetType = ColorType.UINT8)
     buffer.rewind()
     val dataBuffer = DataBuffers.of(buffer)
-    tensor.data().write(dataBuffer)
+    tensor.write(dataBuffer)
 }
 
 
-fun Tensor<TFloat32>.copyTo(colorBuffer: ColorBuffer) {
+
+fun TFloat32.copyTo(colorBuffer: ColorBuffer) {
     val s = shape()
 
     val components = when {
@@ -54,92 +57,84 @@ fun Tensor<TFloat32>.copyTo(colorBuffer: ColorBuffer) {
     val buffer = ByteBuffer.allocateDirect(this.numBytes().toInt())
     buffer.order(ByteOrder.nativeOrder())
     val dataBuffer = DataBuffers.of(buffer.asFloatBuffer())
-    data().read(dataBuffer)
+
+    this.read(dataBuffer)
     buffer.rewind()
     colorBuffer.write(buffer, sourceFormat = format, sourceType = ColorType.FLOAT32)
 }
 
-
+fun TType.summary() {
-fun <T : TType> Tensor<T>.summary() {
+    println("type: ${this.dataType().name}")
-    println("type: ${this.dataType().name()}")
     println("shape: [${this.shape().asArray().joinToString(", ")}]")
 }
 
-fun Tensor<TInt32>.toIntArray(): IntArray {
+fun TInt32.toIntArray(): IntArray {
     val elementCount = this.numBytes() / 4
-    val tensorData = data()
     val targetArray = IntArray(elementCount.toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TInt64>.toLongArray(): LongArray {
+fun TInt64.toLongArray(): LongArray {
     val elementCount = this.numBytes() / 8
-    val tensorData = data()
     val targetArray = LongArray(elementCount.toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TUint8>.toByteArray(): ByteArray {
+fun TUint8.toByteArray(): ByteArray {
     val elementCount = this.numBytes() / 8
-    val tensorData = data()
     val targetArray = ByteArray(elementCount.toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
 
-fun Tensor<TFloat32>.toFloatArray(): FloatArray {
+fun TFloat32.toFloatArray(): FloatArray {
     val elementCount = this.numBytes() / 4
-    val tensorData = data()
     val targetArray = FloatArray(elementCount.toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TFloat32>.toFloatArray2D(): FloatArray2D {
+fun TFloat32.toFloatArray2D(): FloatArray2D {
     val shape = this.shape()
     require(shape.numDimensions() == 2) {
         "tensor has ${shape.numDimensions()} dimensions, need 2"
     }
-    val tensorData = data()
     val targetArray = floatArray2D(shape.size(0).toInt(), shape.size(1).toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TFloat32>.toFloatArray3D(): FloatArray3D {
+fun TFloat32.toFloatArray3D(): FloatArray3D {
     val shape = this.shape()
     require(shape.numDimensions() == 3) {
         "tensor has ${shape.numDimensions()} dimensions, need 3"
     }
-    val tensorData = data()
     val targetArray = floatArray3D(shape.size(0).toInt(), shape.size(1).toInt(), shape.size(2).toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TFloat32>.toFloatArray4D(): FloatArray4D {
+fun TFloat32.toFloatArray4D(): FloatArray4D {
     val shape = this.shape()
     require(shape.numDimensions() == 4) {
         "tensor has ${shape.numDimensions()} dimensions, need 4"
     }
-    val tensorData = data()
     val targetArray = floatArray4D(shape.size(0).toInt(), shape.size(1).toInt(), shape.size(2).toInt(), shape.size(3).toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TFloat64>.toDoubleArray(): DoubleArray {
+fun TFloat64.toDoubleArray(): DoubleArray {
     val elementCount = this.numBytes() / 8
-    val tensorData = data()
     val targetArray = DoubleArray(elementCount.toInt())
-    StdArrays.copyFrom(tensorData, targetArray)
+    StdArrays.copyFrom(this, targetArray)
     return targetArray
 }
 
-fun Tensor<TFloat32>.toColorBuffer(target: ColorBuffer? = null): ColorBuffer {
+fun TFloat32.toColorBuffer(target: ColorBuffer? = null): ColorBuffer {
     val s = shape()
     require(s.numDimensions() == 2 || s.numDimensions() == 3)
 
@@ -168,7 +163,7 @@ fun Tensor<TFloat32>.toColorBuffer(target: ColorBuffer? = null): ColorBuffer {
 
 
 @JvmName("toColorBufferTInt8")
-fun Tensor<TUint8>.toColorBuffer(target: ColorBuffer? = null): ColorBuffer {
+fun TUint8.toColorBuffer(target: ColorBuffer? = null): ColorBuffer {
     val s = shape()
     require(s.numDimensions() == 2 || s.numDimensions() == 3)