Updating extended GAN tests

Signed-off-by: brian <brian@brutex.de>
master
Brian Rosenberger 2023-08-14 14:07:03 +02:00
parent a40d5aa7cf
commit 090c5ab2eb
6 changed files with 374 additions and 119 deletions

View File

@ -36,9 +36,9 @@ import static org.junit.jupiter.api.Assertions.assertTrue;
public class LoadBackendTests { public class LoadBackendTests {
@Test @Test
public void loadBackend() throws ClassNotFoundException, NoSuchFieldException, IllegalAccessException { public void loadBackend() throws NoSuchFieldException, IllegalAccessException {
// check if Nd4j is there // check if Nd4j is there
//Logger.getLogger(LoadBackendTests.class.getName()).info("System java.library.path: " + System.getProperty("java.library.path")); Logger.getLogger(LoadBackendTests.class.getName()).info("System java.library.path: " + System.getProperty("java.library.path"));
final Field sysPathsField = ClassLoader.class.getDeclaredField("sys_paths"); final Field sysPathsField = ClassLoader.class.getDeclaredField("sys_paths");
sysPathsField.setAccessible(true); sysPathsField.setAccessible(true);
sysPathsField.set(null, null); sysPathsField.set(null, null);

View File

@ -37,6 +37,8 @@ import org.datavec.image.loader.NativeImageLoader;
import org.datavec.image.recordreader.ImageRecordReader; import org.datavec.image.recordreader.ImageRecordReader;
import org.datavec.image.transform.*; import org.datavec.image.transform.*;
import org.deeplearning4j.datasets.datavec.RecordReaderDataSetIterator; import org.deeplearning4j.datasets.datavec.RecordReaderDataSetIterator;
import org.deeplearning4j.datasets.iterator.ExistingDataSetIterator;
import org.deeplearning4j.datasets.iterator.INDArrayDataSetIterator;
import org.deeplearning4j.datasets.iterator.impl.MnistDataSetIterator; import org.deeplearning4j.datasets.iterator.impl.MnistDataSetIterator;
import org.deeplearning4j.nn.conf.GradientNormalization; import org.deeplearning4j.nn.conf.GradientNormalization;
import org.deeplearning4j.nn.conf.NeuralNetConfiguration; import org.deeplearning4j.nn.conf.NeuralNetConfiguration;
@ -46,24 +48,27 @@ import org.deeplearning4j.nn.multilayer.MultiLayerNetwork;
import org.deeplearning4j.optimize.listeners.PerformanceListener; import org.deeplearning4j.optimize.listeners.PerformanceListener;
import org.junit.jupiter.api.Tag; import org.junit.jupiter.api.Tag;
import org.junit.jupiter.api.Test; import org.junit.jupiter.api.Test;
import org.nd4j.evaluation.classification.Evaluation;
import org.nd4j.linalg.api.ndarray.INDArray; import org.nd4j.linalg.api.ndarray.INDArray;
import org.nd4j.linalg.dataset.DataSet; import org.nd4j.linalg.dataset.DataSet;
import org.nd4j.linalg.dataset.api.iterator.DataSetIterator; import org.nd4j.linalg.dataset.api.iterator.DataSetIterator;
import org.nd4j.linalg.factory.Nd4j; import org.nd4j.linalg.factory.Nd4j;
import static net.brutex.gan.App2Config.BATCHSIZE;
@Slf4j @Slf4j
public class App2 { public class App2 {
final int INPUT = CHANNELS*DIMENSIONS*DIMENSIONS; final int INPUT = CHANNELS*DIMENSIONS*DIMENSIONS;
static final float COLORSPACE = 255f;
static final int DIMENSIONS = 28; static final int DIMENSIONS = 28;
static final int CHANNELS = 1; static final int CHANNELS = 1;
final int ARRAY_SIZE_PER_SAMPLE = DIMENSIONS*DIMENSIONS*CHANNELS; final int ARRAY_SIZE_PER_SAMPLE = DIMENSIONS*DIMENSIONS*CHANNELS;
final int OUTPUT_PER_PANEL = 10;
final boolean BIAS = true; final boolean BIAS = true;
static final int BATCHSIZE=128;
private JFrame frame2, frame; private JFrame frame2, frame;
static final String OUTPUT_DIR = "d:/out/"; static final String OUTPUT_DIR = "d:/out/";
@ -76,7 +81,7 @@ public class App2 {
Nd4j.getMemoryManager().setAutoGcWindow(15 * 1000); Nd4j.getMemoryManager().setAutoGcWindow(15 * 1000);
MnistDataSetIterator mnistIter = new MnistDataSetIterator(20, 200); MnistDataSetIterator mnistIter = new MnistDataSetIterator(20, 200);
FileSplit fileSplit = new FileSplit(new File("c:/users/brian/downloads/humans2"), NativeImageLoader.getALLOWED_FORMATS()); FileSplit fileSplit = new FileSplit(new File("c:/users/brian/downloads/humans3"), NativeImageLoader.getALLOWED_FORMATS());
ImageTransform transform = new ColorConversionTransform(new Random(42), 7 ); ImageTransform transform = new ColorConversionTransform(new Random(42), 7 );
ImageTransform transform2 = new ShowImageTransform("Tester", 30); ImageTransform transform2 = new ShowImageTransform("Tester", 30);
ImageTransform transform3 = new ResizeImageTransform(DIMENSIONS, DIMENSIONS); ImageTransform transform3 = new ResizeImageTransform(DIMENSIONS, DIMENSIONS);
@ -129,12 +134,94 @@ public class App2 {
log.info("Generator Summary:\n{}", gen.summary()); log.info("Generator Summary:\n{}", gen.summary());
log.info("GAN Summary:\n{}", gan.summary()); log.info("GAN Summary:\n{}", gan.summary());
dis.addTrainingListeners(new PerformanceListener(10, true, "DIS")); dis.addTrainingListeners(new PerformanceListener(3, true, "DIS"));
gen.addTrainingListeners(new PerformanceListener(10, true, "GEN")); //gen.addTrainingListeners(new PerformanceListener(3, true, "GEN")); //is never trained separately from GAN
gan.addTrainingListeners(new PerformanceListener(10, true, "GAN")); gan.addTrainingListeners(new PerformanceListener(3, true, "GAN"));
/*
Thread vt =
new Thread(
new Runnable() {
@Override
public void run() {
while (true) {
visualize(0, 0, gen);
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
throw new RuntimeException(e);
}
}
}
});
vt.start();
*/
int j = 0; App2Display display = new App2Display();
for (int i = 0; i < 51; i++) { //epoch //Repack training data with new fake/real label. Original MNist has 10 labels, one for each digit
DataSet data = null;
int j =0;
for(int i=0;i<App2Config.EPOCHS;i++) {
log.info("Epoch {}", i);
data = new DataSet(Nd4j.rand(BATCHSIZE, 784), label_fake);
while (trainData.hasNext()) {
j++;
INDArray real = trainData.next().getFeatures();
INDArray fakeIn = Nd4j.rand(BATCHSIZE, App2Config.INPUT);
INDArray fake = gan.activateSelectedLayers(0, gen.getLayers().length - 1,
Nd4j.rand(BATCHSIZE, App2Config.INPUT));
//sigmoid output is -1 to 1
fake.addi(1f).divi(2f);
if (j % 50 == 1) {
display.visualize(new INDArray[] {fake}, App2Config.OUTPUT_PER_PANEL, false);
display.visualize(new INDArray[] {real}, App2Config.OUTPUT_PER_PANEL, true);
}
DataSet realSet = new DataSet(real, label_real);
DataSet fakeSet = new DataSet(fake, label_fake);
//start next round if there are not enough images left to have a full batchsize dataset
if(real.length() < ARRAY_SIZE_PER_SAMPLE*BATCHSIZE) {
log.warn("Your total number of input images is not a multiple of {}, "
+ "thus skipping {} images to make it fit", BATCHSIZE, real.length()/ARRAY_SIZE_PER_SAMPLE);
break;
}
//if(real.length()/BATCHSIZE!=784) break;
data = DataSet.merge(Arrays.asList(data, realSet, fakeSet));
}
//fit the discriminator
dis.fit(data);
dis.fit(data);
// Update the discriminator in the GAN network
updateGan(gen, dis, gan);
//reset the training data and fit the complete GAN
if (trainData.resetSupported()) {
trainData.reset();
} else {
log.error("Trainingdata {} does not support reset.", trainData.toString());
}
gan.fit(new DataSet(Nd4j.rand(BATCHSIZE, App2Config.INPUT), label_real));
if (trainData.resetSupported()) {
trainData.reset();
} else {
log.error("Trainingdata {} does not support reset.", trainData.toString());
}
log.info("Updated GAN's generator from gen.");
updateGen(gen, gan);
gen.save(new File("mnist-mlp-generator.dlj"));
}
//vt.stop();
/*
int j;
for (int i = 0; i < App2Config.EPOCHS; i++) { //epoch
j=0;
while (trainData.hasNext()) { while (trainData.hasNext()) {
j++; j++;
DataSet next = trainData.next(); DataSet next = trainData.next();
@ -212,122 +299,25 @@ public class App2 {
log.info("Updated GAN's generator from gen."); log.info("Updated GAN's generator from gen.");
gen.save(new File("mnist-mlp-generator.dlj")); gen.save(new File("mnist-mlp-generator.dlj"));
} }
} */
private static JFrame visualize(INDArray[] samples, int batchElements, JFrame frame, boolean isOrig) {
if (isOrig) {
frame.setTitle("Viz Original");
} else {
frame.setTitle("Generated");
}
frame.setDefaultCloseOperation(WindowConstants.DISPOSE_ON_CLOSE);
frame.setLayout(new BorderLayout());
JPanel panelx = new JPanel();
panelx.setLayout(new GridLayout(4, 4, 8, 8));
for (INDArray sample : samples) {
for(int i = 0; i<batchElements; i++) {
panelx.add(getImage(sample, i, isOrig));
}
}
frame.add(panelx, BorderLayout.CENTER);
frame.setVisible(true);
frame.revalidate();
frame.setMinimumSize(new Dimension(300, 20));
frame.pack();
return frame;
}
private static JLabel getImage(INDArray tensor, int batchElement, boolean isOrig) {
final BufferedImage bi;
if(CHANNELS >1) {
bi = new BufferedImage(DIMENSIONS, DIMENSIONS, BufferedImage.TYPE_INT_RGB); //need to change here based on channels
} else {
bi = new BufferedImage(DIMENSIONS, DIMENSIONS, BufferedImage.TYPE_BYTE_GRAY); //need to change here based on channels
}
final int imageSize = DIMENSIONS * DIMENSIONS;
final int offset = batchElement * imageSize;
int pxl = offset * CHANNELS; //where to start in the INDArray
//Image in NCHW - channels first format
for (int c = 0; c < CHANNELS; c++) { //step through the num channels for each pixel
for (int y = 0; y < DIMENSIONS; y++) { // step through the columns x
for (int x = 0; x < DIMENSIONS; x++) { //step through the rows y
float f_pxl = tensor.getFloat(pxl) * COLORSPACE;
if(isOrig) log.trace("'{}.' Image (x,y,c): ({}, {}, {}) with INDArray with index {} and value '{}'", batchElement, x, y, c, pxl, f_pxl);
bi.getRaster().setSample(x, y, c, f_pxl);
pxl++; //next item in INDArray
}
}
}
ImageIcon orig = new ImageIcon(bi);
Image imageScaled = orig.getImage().getScaledInstance((4 * DIMENSIONS), (4 * DIMENSIONS), Image.SCALE_DEFAULT);
ImageIcon scaled = new ImageIcon(imageScaled);
if(! isOrig) saveImage(imageScaled, batchElement, isOrig);
return new JLabel(scaled);
} }
private static void saveImage(Image image, int batchElement, boolean isOrig) {
String outputDirectory = OUTPUT_DIR; // Set the output directory where the images will be saved
try {
// Save the images to disk
saveImage(image, outputDirectory, UUID.randomUUID().toString()+".png");
log.debug("Images saved successfully.");
} catch (IOException e) {
log.error("Error saving the images: {}", e.getMessage());
}
}
private static void saveImage(Image image, String outputDirectory, String fileName) throws IOException {
File directory = new File(outputDirectory);
if (!directory.exists()) {
directory.mkdir();
}
File outputFile = new File(directory, fileName);
ImageIO.write(imageToBufferedImage(image), "png", outputFile);
}
public static BufferedImage imageToBufferedImage(Image image) {
if (image instanceof BufferedImage) {
return (BufferedImage) image;
}
// Create a buffered image with the same dimensions and transparency as the original image
BufferedImage bufferedImage;
if (CHANNELS > 1) {
bufferedImage =
new BufferedImage(
image.getWidth(null), image.getHeight(null), BufferedImage.TYPE_INT_ARGB);
} else {
bufferedImage =
new BufferedImage(
image.getWidth(null), image.getHeight(null), BufferedImage.TYPE_BYTE_GRAY);
}
// Draw the original image onto the buffered image
Graphics2D g2d = bufferedImage.createGraphics();
g2d.drawImage(image, 0, 0, null);
g2d.dispose();
return bufferedImage;
}
private static void updateGen(MultiLayerNetwork gen, MultiLayerNetwork gan) { private static void updateGen(MultiLayerNetwork gen, MultiLayerNetwork gan) {
for (int i = 0; i < gen.getLayers().length; i++) { for (int i = 0; i < gen.getLayers().length; i++) {
gen.getLayer(i).setParams(gan.getLayer(i).getParams()); gen.getLayer(i).setParams(gan.getLayer(i).getParams());
@ -341,4 +331,41 @@ public class App2 {
} }
} }
@Test
void testDiskriminator() throws IOException {
MultiLayerNetwork net = new MultiLayerNetwork(App2Config.discriminator());
net.init();
net.addTrainingListeners(new PerformanceListener(10, true, "DIS"));
DataSetIterator trainData = new MnistDataSetIterator(BATCHSIZE, true, 42);
DataSet data = null;
for(int i=0;i<App2Config.EPOCHS;i++) {
log.info("Epoch {}", i);
data = new DataSet(Nd4j.rand(BATCHSIZE, 784), label_fake);
while (trainData.hasNext()) {
INDArray real = trainData.next().getFeatures();
long[] l = new long[]{BATCHSIZE, real.length() / BATCHSIZE};
INDArray fake = Nd4j.rand(l );
DataSet realSet = new DataSet(real, label_real);
DataSet fakeSet = new DataSet(fake, label_fake);
if(real.length()/BATCHSIZE!=784) break;
data = DataSet.merge(Arrays.asList(data, realSet, fakeSet));
}
net.fit(data);
trainData.reset();
}
long[] l = new long[]{BATCHSIZE, 784};
INDArray fake = Nd4j.rand(l );
DataSet fakeSet = new DataSet(fake, label_fake);
data = DataSet.merge(Arrays.asList(data, fakeSet));
ExistingDataSetIterator iter = new ExistingDataSetIterator(data);
Evaluation eval = net.evaluate(iter);
log.info( "\n" + eval.confusionMatrix());
}
} }

View File

@ -36,10 +36,17 @@ import org.nd4j.linalg.lossfunctions.LossFunctions;
public class App2Config { public class App2Config {
public static final int INPUT = 100; public static final int INPUT = 100;
public static final int BATCHSIZE=150;
public static final int X_DIM = 28; public static final int X_DIM = 28;
public static final int y_DIM = 28; public static final int Y_DIM = 28;
public static final int CHANNELS = 1; public static final int CHANNELS = 1;
public static final int EPOCHS = 50;
public static final IUpdater UPDATER = Adam.builder().learningRate(0.0002).beta1(0.5).build(); public static final IUpdater UPDATER = Adam.builder().learningRate(0.0002).beta1(0.5).build();
public static final IUpdater UPDATER_DIS = Adam.builder().learningRate(0.02).beta1(0.5).build();
public static final boolean SHOW_GENERATED = true;
public static final float COLORSPACE = 255f;
final static int OUTPUT_PER_PANEL = 10;
static LayerConfiguration[] genLayerConfig() { static LayerConfiguration[] genLayerConfig() {
return new LayerConfiguration[] { return new LayerConfiguration[] {
@ -158,7 +165,7 @@ public class App2Config {
.gradientNormalization(GradientNormalization.RenormalizeL2PerLayer) .gradientNormalization(GradientNormalization.RenormalizeL2PerLayer)
.gradientNormalizationThreshold(100) .gradientNormalizationThreshold(100)
.seed(42) .seed(42)
.updater(UPDATER) .updater(UPDATER_DIS)
.weightInit(WeightInit.XAVIER) .weightInit(WeightInit.XAVIER)
// .weightNoise(new WeightNoise(new NormalDistribution(0.5, 0.5))) // .weightNoise(new WeightNoise(new NormalDistribution(0.5, 0.5)))
.weightNoise(null) .weightNoise(null)

View File

@ -0,0 +1,160 @@
/*
*
* ******************************************************************************
* *
* * This program and the accompanying materials are made available under the
* * terms of the Apache License, Version 2.0 which is available at
* * https://www.apache.org/licenses/LICENSE-2.0.
* *
* * See the NOTICE file distributed with this work for additional
* * information regarding copyright ownership.
* * Unless required by applicable law or agreed to in writing, software
* * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* * License for the specific language governing permissions and limitations
* * under the License.
* *
* * SPDX-License-Identifier: Apache-2.0
* *****************************************************************************
*
*/
package net.brutex.gan;
import com.google.inject.Singleton;
import lombok.extern.slf4j.Slf4j;
import org.nd4j.linalg.api.ndarray.INDArray;
import javax.imageio.ImageIO;
import javax.swing.*;
import java.awt.*;
import java.awt.color.ColorSpace;
import java.awt.image.BufferedImage;
import java.io.File;
import java.io.IOException;
import java.util.UUID;
import static net.brutex.gan.App2.OUTPUT_DIR;
import static net.brutex.gan.App2Config.*;
@Slf4j
@Singleton
public class App2Display {
private final JFrame frame = new JFrame();
private final App2GUI display = new App2GUI();
private final JPanel real_panel;
private final JPanel fake_panel;
public App2Display() {
frame.setDefaultCloseOperation(WindowConstants.DISPOSE_ON_CLOSE);
frame.setContentPane(display.getOverall_panel());
frame.setMinimumSize(new Dimension(300, 20));
frame.pack();
frame.setVisible(true);
real_panel = display.getReal_panel();
fake_panel = display.getGen_panel();
real_panel.setLayout(new GridLayout(4, 4, 8, 8));
fake_panel.setLayout(new GridLayout(4, 4, 8, 8));
}
public void visualize(INDArray[] samples, int batchElements, boolean isOrig) {
for (INDArray sample : samples) {
for(int i = 0; i<batchElements; i++) {
final Image img = this.getImage(sample, i, isOrig);
final ImageIcon icon = new ImageIcon(img);
if(isOrig) {
if(real_panel.getComponents().length>=OUTPUT_PER_PANEL) {
real_panel.remove(0);
}
real_panel.add(new JLabel(icon));
} else {
if(fake_panel.getComponents().length>=OUTPUT_PER_PANEL) {
fake_panel.remove(0);
}
fake_panel.add(new JLabel(icon));
}
}
}
frame.pack();
frame.repaint();
}
public Image getImage(INDArray tensor, int batchElement, boolean isOrig) {
final BufferedImage bi;
if(CHANNELS >1) {
bi = new BufferedImage(X_DIM, Y_DIM, BufferedImage.TYPE_INT_RGB); //need to change here based on channels
} else {
bi = new BufferedImage(X_DIM, Y_DIM, BufferedImage.TYPE_BYTE_GRAY); //need to change here based on channels
}
final int imageSize = X_DIM * Y_DIM;
final int offset = batchElement * imageSize;
int pxl = offset * CHANNELS; //where to start in the INDArray
//Image in NCHW - channels first format
for (int c = 0; c < CHANNELS; c++) { //step through the num channels for each pixel
for (int y = 0; y < X_DIM; y++) { // step through the columns x
for (int x = 0; x < Y_DIM; x++) { //step through the rows y
float f_pxl = tensor.getFloat(pxl) * COLORSPACE;
if(isOrig) log.trace("'{}.'{} Image (x,y,c): ({}, {}, {}) with INDArray with index {} and value '{}'", batchElement, isOrig ? "Real" : "Fake", x, y, c, pxl, f_pxl);
bi.getRaster().setSample(x, y, c, f_pxl);
pxl++; //next item in INDArray
}
}
}
ImageIcon orig = new ImageIcon(bi);
Image imageScaled = orig.getImage().getScaledInstance((4 * X_DIM), (4 * Y_DIM), Image.SCALE_DEFAULT);
ImageIcon scaled = new ImageIcon(imageScaled);
//if(! isOrig) saveImage(imageScaled, batchElement, isOrig);
return imageScaled;
}
private static void saveImage(Image image, int batchElement, boolean isOrig) {
String outputDirectory = OUTPUT_DIR; // Set the output directory where the images will be saved
try {
// Save the images to disk
saveImage(image, outputDirectory, UUID.randomUUID().toString()+".png");
log.debug("Images saved successfully.");
} catch (IOException e) {
log.error("Error saving the images: {}", e.getMessage());
}
}
private static void saveImage(Image image, String outputDirectory, String fileName) throws IOException {
File directory = new File(outputDirectory);
if (!directory.exists()) {
directory.mkdir();
}
File outputFile = new File(directory, fileName);
ImageIO.write(imageToBufferedImage(image), "png", outputFile);
}
public static BufferedImage imageToBufferedImage(Image image) {
if (image instanceof BufferedImage) {
return (BufferedImage) image;
}
// Create a buffered image with the same dimensions and transparency as the original image
BufferedImage bufferedImage;
if (CHANNELS > 1) {
bufferedImage =
new BufferedImage(
image.getWidth(null), image.getHeight(null), BufferedImage.TYPE_INT_ARGB);
} else {
bufferedImage =
new BufferedImage(
image.getWidth(null), image.getHeight(null), BufferedImage.TYPE_BYTE_GRAY);
}
// Draw the original image onto the buffered image
Graphics2D g2d = bufferedImage.createGraphics();
g2d.drawImage(image, 0, 0, null);
g2d.dispose();
return bufferedImage;
}
}

View File

@ -0,0 +1,61 @@
package net.brutex.gan;
import javax.swing.JPanel;
import javax.swing.JSplitPane;
import javax.swing.JLabel;
import java.awt.BorderLayout;
public class App2GUI extends JPanel {
/**
*
*/
private static final long serialVersionUID = 1L;
private JPanel overall_panel;
private JPanel real_panel;
private JPanel gen_panel;
/**
* Create the panel.
*/
public App2GUI() {
overall_panel = new JPanel();
add(overall_panel);
JSplitPane splitPane = new JSplitPane();
overall_panel.add(splitPane);
JPanel p1 = new JPanel();
splitPane.setLeftComponent(p1);
p1.setLayout(new BorderLayout(0, 0));
JLabel lblNewLabel = new JLabel("Generator");
p1.add(lblNewLabel, BorderLayout.NORTH);
gen_panel = new JPanel();
p1.add(gen_panel, BorderLayout.SOUTH);
JPanel p2 = new JPanel();
splitPane.setRightComponent(p2);
p2.setLayout(new BorderLayout(0, 0));
JLabel lblNewLabel_1 = new JLabel("Real");
p2.add(lblNewLabel_1, BorderLayout.NORTH);
real_panel = new JPanel();
p2.add(real_panel, BorderLayout.SOUTH);
}
public JPanel getOverall_panel() {
return overall_panel;
}
public JPanel getReal_panel() {
return real_panel;
}
public JPanel getGen_panel() {
return gen_panel;
}
}