Details

This PR adds JNI wrappers to the OpenVINO Java API and includes a hello_query_device sample that demonstrates the usage of the Query Device API feature to print device configuration and default config values.

Below is a list of JNI wrappers added:

• Read and compile a model directly from the model path

  ov::Core	Core
  compile_model(model_path)	compile_model(modelPath)
  compile_model(model_path, device_name)	compile_model(modelPath, deviceName)
  compile_model(model_path, device_name, properties={})	compile_model(modelPath, deviceName, properties)

• Compile an ov.Model with the given properties

  ov::Core	Core
  compile_model(model, device_name, properties={})	compile_model(model, deviceName, properties)

• List devices available for inference

  ov::Core	Core
  get_available_devices()	get_available_devices()

• Extract data from ov.Any type

  ov::Any	Any
  	asString()
  	asList()

• List the input and output tensor information of a CompiledModel. Returns the tensor shape, name and element type of each input/output tensor.

  ov::CompiledModel	CompiledModel
  inputs()	inputs()
  outputs()	outputs()

• Set input/output tensors to infer on by tensor name when the model has multiple input/output tensors

  ov::InferRequest	InferRequest
  set_tensor(tensor_name, tensor)	set_tensor(tensorName, tensor)

• Create integer tensors and extract data as an int array

  ov::Tensor	Tensor
  Tensor(shape, data)	Tensor(shape, data)
  data()	as_int()

• Create Long tensors

  ov::Tensor	Tensor
  Tensor(shape, data)	Tensor(shape, data)

• Save model into IR format from memory

  ov::serialize(model, xml_path, bin_path)	Openvino.serialize(model, xmlPath, binPath)