如何读取 TensorBuffer

如何解决如何读取 TensorBuffer

我是 Tensorflow 和 Android 的新手，我正在尝试对来自外部加速度计的数据进行分类，以预测它与左、右、前、后或中心对齐的天气。所以我使用colab训练模型并将其转换为tflite并将其添加到Android App中，我无法理解输出

package com.yogai.tensorflowlava;

import androidx.appcompat.app.AppCompatActivity;

import android.content.Context;
import android.os.Bundle;
import android.util.Log;
import android.view.View;
import android.widget.Button;
import android.widget.EditText;
import android.widget.Toast;
import android.widget.Toolbar;

import com.yogai.tensorflowlava.ml.Adxl345;

import org.tensorflow.lite.DataType;
import org.tensorflow.lite.TensorFlowLite;
import org.tensorflow.lite.support.tensorbuffer.TensorBuffer;

import java.io.IOException;
import java.nio.ByteBuffer;

public class MainActivity extends AppCompatActivity {
        Button submitButton;
        EditText editText;
        String text;
        String TAG = "My_APp";
   // private Context context;

    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
        Context context;

        editText = findViewById(R.id.editText);
        submitButton = findViewById(R.id.submitButton);
        submitButton.setonClickListener(new View.OnClickListener() {
            @Override
            public void onClick(View view) {

                text = editText.getText().toString().trim();
                //String str = "geekss@for@geekss";
//                String[] arrOfStr = text.split(",",3);
//
//               // String[] strings = new String[] {"1","2","3","4"};
//                if (android.os.Build.VERSION.SDK_INT >= android.os.Build.VERSION_CODES.N) {
//                    Float[] floats = Arrays.stream(arrOfStr).map(Float::valueOf).toArray(Float[]::new);
//                }
                String[] parts = text.split(",");
                float[] numbers = new float[parts.length];
                for (int i = 0; i < parts.length; ++i) {
                    float number = Float.parseFloat(parts[i]);
                    float rounded = (int) Math.round(number * 1000) / 1000f;
                    numbers[i] = rounded;
                }


//                float[][] array2d = new float[1][3];
//
//
//
//                for(int j=1;j<4;j++) {
//                    array2d[1][j] = numbers[j];
//
//
//                }
//                Float testValue = array2d[1][1]+1;
                Log.d(TAG,String.valueOf(numbers[1]));
                ByteBuffer.allocate(4).putFloat(numbers[0]).array();
                byte[] byteArray= FloatArray2ByteArray(numbers);
                ByteBuffer byteBuffer = ByteBuffer.wrap(byteArray);
                getoutput(byteBuffer);






            }
        });
    }


        public static byte[] FloatArray2ByteArray(float[] values){
            ByteBuffer buffer = ByteBuffer.allocate(4 * values.length);

            for (float value : values){
                buffer.putFloat(value);
            }

            return buffer.array();
        }



    private void getoutput(ByteBuffer byteBuffer) {



        try {
            Adxl345 model = Adxl345.newInstance(getApplicationContext());

            // Creates inputs for reference.
            TensorBuffer inputFeature0 = TensorBuffer.createFixedSize(new int[]{1,3},DataType.FLOAT32);
            inputFeature0.loadBuffer(byteBuffer);

            // Runs model inference and gets result.
            Adxl345.Outputs outputs = model.process(inputFeature0);
            TensorBuffer outputFeature0 = outputs.getoutputFeature0AsTensorBuffer();

            String converted = new String(buffer.array(),"UTF-8");
            
            Toast.makeText(this,"output: "+outputFeature0.toString(),Toast.LENGTH_SHORT).show();

            // Releases model resources if no longer used.
            model.close();
        } catch (IOException e) {
            // Todo Handle the exception
            Toast.makeText(this,"Error: "+ e.toString(),Toast.LENGTH_SHORT).show();
        }
    }




}