Version Matrix / java.lang.IncompatibleClassChangeError

[sauterl]

First of all, thanks for the great work and great examples!

I tried to re-create the object detection example in an existing project (OpenJDK 17, Windows 10) and used following setup (basically got all the latest version):

/** Deep Java Library*/
    implementation group: 'ai.djl', name: 'api', version: '0.18.0'
    implementation group: 'ai.djl', name: 'repository', version: '0.4.1'
    runtimeOnly group: 'ai.djl.pytorch', name: 'pytorch-native-auto', version: '1.9.1'
    runtimeOnly group: 'ai.djl.pytorch', name: 'pytorch-model-zoo', version: '0.18.0'

However, I do get the following exception:

Exception in thread "main" java.lang.IncompatibleClassChangeError: class ai.djl.pytorch.zoo.PtModelZoo has interface ai.djl.repository.zoo.ModelZoo as super class
	at java.base/java.lang.ClassLoader.defineClass1(Native Method)
	at java.base/java.lang.ClassLoader.defineClass(ClassLoader.java:1012)
	at java.base/java.security.SecureClassLoader.defineClass(SecureClassLoader.java:150)
	at java.base/jdk.internal.loader.BuiltinClassLoader.defineClass(BuiltinClassLoader.java:862)
	at java.base/jdk.internal.loader.BuiltinClassLoader.findClassOnClassPathOrNull(BuiltinClassLoader.java:760)
	at java.base/jdk.internal.loader.BuiltinClassLoader.loadClassOrNull(BuiltinClassLoader.java:681)
	at java.base/jdk.internal.loader.BuiltinClassLoader.loadClass(BuiltinClassLoader.java:639)
	at java.base/jdk.internal.loader.ClassLoaders$AppClassLoader.loadClass(ClassLoaders.java:188)
	at java.base/java.lang.ClassLoader.loadClass(ClassLoader.java:520)
	at ai.djl.pytorch.zoo.PtZooProvider.getModelZoo(PtZooProvider.java:26)
	at ai.djl.repository.zoo.ModelZoo.loadModel(ModelZoo.java:99)

which indicates, that there is somewhere a version mismatch. Yet another peculiar consequence is, that using the above version-interplay leads to having the model being loaded via ModelZoo.loadModel instead of the example's criteria.loadModel()

My sandbox code:

package org.vitrivr.cineast.core;

import ai.djl.Application;
import ai.djl.ModelException;
import ai.djl.engine.Engine;
import ai.djl.inference.Predictor;
import ai.djl.modality.cv.Image;
import ai.djl.modality.cv.ImageFactory;
import ai.djl.modality.cv.output.DetectedObjects;
import ai.djl.repository.zoo.Criteria;
import ai.djl.repository.zoo.ModelZoo;
import ai.djl.repository.zoo.ZooModel;
import ai.djl.training.util.ProgressBar;
import ai.djl.translate.TranslateException;
import java.io.IOException;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.Paths;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public class Sandbox {
  private static final Logger logger = LoggerFactory.getLogger(Sandbox.class);


  public static void main(String[] args) throws IOException, ModelException, TranslateException {
    DetectedObjects detection = predict();
    logger.info("{}", detection);
  }

  public static DetectedObjects predict() throws IOException, ModelException, TranslateException {
    String url = "https://github.com/awslabs/djl/raw/master/examples/src/test/resources/dog_bike_car.jpg";
    Image img = ImageFactory.getInstance().fromUrl(url);

    String backbone;
    if ("TensorFlow".equals(Engine.getDefaultEngineName())) {
      backbone = "mobilenet_v2";
    } else {
      backbone = "resnet50";
    }

    Criteria<Image, DetectedObjects> criteria =
        Criteria.builder()
            .optApplication(Application.CV.OBJECT_DETECTION)
            .setTypes(Image.class, DetectedObjects.class)
            .optFilter("backbone", backbone)
            .optEngine(Engine.getDefaultEngineName())
            .optProgress(new ProgressBar())
            .build();

    try (ZooModel<Image, DetectedObjects> model = ModelZoo.loadModel(criteria)) {
      try (Predictor<Image, DetectedObjects> predictor = model.newPredictor()) {
        DetectedObjects detection = predictor.predict(img);
        saveBoundingBoxImage(img, detection);
        return detection;
      }
    }
  }

  private static void saveBoundingBoxImage(Image img, DetectedObjects detection)
      throws IOException {
    Path outputDir = Paths.get("build/output");
    Files.createDirectories(outputDir);

    img.drawBoundingBoxes(detection);

    Path imagePath = outputDir.resolve("detected-dog_bike_car.png");
    // OpenJDK can't save jpg with alpha channel
    img.save(Files.newOutputStream(imagePath), "png");
    logger.info("Detected objects image has been saved in: {}", imagePath);
  }
}

Can anyone with more experience / knowledge in this domain point me to the version mismatch? Cloning the djl repository and running the example from scratch works without a problem.

Thanks in advance!

[frankfliu]

We strongly recommend you to use DJL's BOM. It ensure all the versions are matching with each other.

Both ai.djl:repository and ai.djl.pytorch:pytorch-native-auto are deprecated. Please remove them from the dependency.

You can find deprecated package information here: https://docs.djl.ai/master/docs/development/dependency_management.html

[sauterl]

Great! Thanks for the fast reply - this solved the issue.

For anyone landing via google here:

Replacing the OP gradle dependency lines solves the issue:

/** Deep Java Library*/
    implementation platform("ai.djl:bom:0.18.0")
    implementation "ai.djl.pytorch:pytorch-model-zoo" // No version required