Question

当我在Android中使用tensorflow-lite时，更改正式支持的模型后，发生了错误

我模型的输入有两个输入，分别是32 * 128 * 128 * 1，我模仿了输入的方式，但是失败了

这是我项目中的一些代码

  @Override
  protected void processImage() {
    try {
      float[][][][] stft=readTxt("file:///android_asset/stft.txt");
      float[][][][] mfcc=readTxt("file:///android_asset/mfcc.txt");
      final List<Classifier.Recognition> results = detector.recognizeImage(stft, mfcc);

    }
    catch(Exception e)
    {
      e.printStackTrace();
      System.out.print("can't get Bitmap");
    }

  }

  @Override
  public List<Recognition> recognizeImage(final float[][][][] stft,final float[][][][] mfcc) {
    // Log this method so that it can be analyzed with systrace.
    Trace.beginSection("recognizeImage");

    Trace.beginSection("preprocessBitmap");
    // Preprocess the image data from 0-255 int to normalized float based
    // on the provided parameters.
    LOGGER.i("begin run model");
    Trace.endSection(); // preprocessBitmap

    // Copy the input data into TensorFlow.
    Trace.beginSection("feed");
    imgDatam.rewind();
    imgDatas.rewind();
    for(int i=0; i<32; i++){
      for(int j=0;j<128;j++){
        for(int k =0;k<128;k++){
          imgDatam.putFloat(mfcc[i][j][k][0]);
          imgDatas.putFloat(stft[i][j][k][0]);
        }

      }

    }

    outputClasses = new float[32][NUM_DETECTIONS];

    Object[] inputArray = {imgDatas,imgDatam};

    Map<Integer, Object> outputMap = new HashMap<>();

    outputMap.put(0, outputClasses);

    Trace.endSection();

    // Run the inference call.
    Trace.beginSection("run");
    try{
      tfLite.runForMultipleInputsOutputs(inputArray, outputMap);
    }
    catch(Exception e)
    {
      e.printStackTrace();
    }
    LOGGER.i("finished");
    Trace.endSection();

    // Show the best detections.
    // after scaling them back to the input size.
    final ArrayList<Recognition> recognitions = new ArrayList<>(NUM_DETECTIONS);

    Trace.endSection(); // "recognizeImage"
    return recognitions;
  }

错误显示：

org.tensorflow.lite.examples.detection A / libc：致命信号6（SIGABRT），代码-6（SI_TKILL）在tid 23610（mpls.detection），pid 23610（mples.detection）中？ A / DEBUG：pid：23610，tid：23610，名称：mples.detection >>> org.tensorflow.lite.examples.detection <<< ？ A /调试：＃01 pc 00000000001220b4 /data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag==/lib/arm64/libtensorflowlite_jni.so ？ A /调试：＃02 pc 00000000001216e4 /data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag==/lib/arm64/libtensorflowlite_jni.so ？ A /调试：＃03 pc 00000000001624ac /data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag==/lib/arm64/libtensorflowlite_jni.so ？ A /调试：＃04 pc 0000000000165974 /data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag==/lib/arm64/libtensorflowlite_jni.so ？ A /调试：＃05 pc 000000000000ed10 /data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag==/lib/arm64/libtensorflowlite_jni.so（Java_org_tensorflow_lite_NativeInterpreterWrapper_run + 32）？ A /调试：＃13 pc 00000000001770ee /dev/ashmem/dalvik-classes.dex从/data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag == / split_lib_dependencies_apk.apk（已删除）在内存中提取org.tensorflow.lite.NativeInterpreterWrapper.run + 158）？ A /调试：＃19 pc 00000000001768f6 /dev/ashmem/dalvik-classes.dex从/data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag == / split_lib_dependencies_apk.apk（已删除）在内存中提取org.tensorflow.lite.Interpreter.runForMultipleInputsOutputs + 10）？ A /调试：＃23 pc 0000000000007234 / dev / ashmem / dalvik-jit-code-cache（已删除）（org.tensorflow.lite.examples.detection.tflite.TFLiteObjectDetectionAPIModel.recognizeImage + 1428）？ A /调试：＃28 pc 000000000000200e /dev/ashmem/dalvik-classes.dex从/data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag == / split_lib_slice_4_apk.apk（已删除）在内存中提取org.tensorflow.lite.examples.detection.tflite.TFLiteObjectDetectionAPIModel.recognizeImage + 206）？ A /调试：＃34 pc 000000000001d408 /dev/ashmem/dalvik-classes.dex从/data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag == / split_lib_slice_7_apk.apk（已删除）在内存中提取org.tensorflow.lite.examples.detection.DetectorActivity.processImage + 60）？ A /调试：＃40 pc 0000000000015c6a /dev/ashmem/dalvik-classes.dex从/data/app/org.tensorflow.lite.examples.detection-kKVx47dAvHYDUZmU-_Wkag == / split_lib_slice_7_apk.apk（已删除）在内存中提取org.tensorflow.lite.examples.detection.CameraActivity.onCreate + 174）？ E / adbd：无法连接至套接字'localabstract：org.tensorflow.lite.examples.detection'：连接被拒绝

无论如何，谢谢您的帮助，我知道这可能很难，因为我什至找不到关于tensorflow lite的指南

如何在Android中使用Tensorflow时发生错误

0 个答案: