如何为Tensorflow模型服务器创建Java客户端

<dependencies>
    <dependency>
        <groupId>org.tensorflow</groupId>
        <artifactId>tensorflow</artifactId>
        <version>2.7.0</version>
    </dependency>
</dependencies>

String serverAddress = "localhost";
int serverPort = 9000;

Socket socket = new Socket(serverAddress, serverPort);

import org.tensorflow.Tensor;
import org.tensorflow.TensorFlow;

float[] inputData = {1.0f, 2.0f, 3.0f, 4.0f};
long[] inputShape = {1, 4};

Tensor inputTensor = Tensor.create(inputShape, FloatBuffer.wrap(inputData));

OutputStream outputStream = socket.getOutputStream();
InputStream inputStream = socket.getInputStream();

// 发送请求数据
byte[] inputDataBytes = inputTensor.bytesValue();
outputStream.write(inputDataBytes);
outputStream.flush();

// 接收响应数据
byte[] responseDataBytes = new byte[1024];
int bytesRead = inputStream.read(responseDataBytes);
byte[] responseData = Arrays.copyOf(responseDataBytes, bytesRead);

// 解析响应数据
Tensor outputTensor = Tensor.create(FloatBuffer.wrap(responseData));
float[] outputData = new float[outputTensor.numElements()];
outputTensor.copyTo(outputData);

System.out.println("预测结果：");
for (float output : outputData) {
    System.out.println(output);
}