The Deep Java Library (DJL) model zoo contains engine-agnostic models. All the models have a built-in Translator and can be used for inference out of the box.
You can find general ModelZoo and model loading document here:
The latest javadocs can be found on the djl.ai website.
You can also build the latest javadocs locally using the following command:
# for Linux/macOS: ./gradlew javadoc # for Windows: ..\gradlew javadoc
The javadocs output is built in the build/doc/javadoc folder.
You can pull the model zoo from the central Maven repository by including the following dependency in your
<dependency> <groupId>ai.djl</groupId> <artifactId>model-zoo</artifactId> <version>0.8.0</version> </dependency>
You can find the Multilayer Perceptrons (MLP) and Resnet50 pre-trained models in the model zoo.
How to find a pre-trained model in the model zoo¶
In a model zoo repository, there can be many pre-trained models that belong to the same model family.
You can use the
ModelZoo class to search for the model you need.
First, decide which model family you want to use. Then, define your key/values search criteria
to narrow down the model you want. If there are multiple models that match your search criteria, the first
model found is returned. A ModelNotFoundException will be thrown if no matching model is found.
The following is an example of the criteria to find a Resnet50-v1 model that has been trained on the imagenet dataset:
Criteria<Image, Classification> criteria = Criteria.builder() .optApplication(Application.CV.OBJECT_DETECTION) .setTypes(Image.class, Classification.class) .optFilter("layer", "50") .optFilter("flavor", "v1") .optFilter("dataset", "cifar10") .build(); ZooModel<Image, Classification> ssd = ModelZoo.loadModel(criteria));
If you already known which
ModelLoader to use, you can simply do the following:
Map<String, String> filter = new HashMap<>(); filter.put("layers", "50"); filter.put("flavor", "v1"); filter.put("dataset", "cifar10"); ZooModel<Image, Classification> model = BasicModelZoo.RESNET.loadModel(filter);