Question

我已经读过here，现在tensorflow lite可以从tensorflow和keras LSTM模型进行转换。

在转换模型之前，一切正常，并且我有了model.tflite文件。我已经尝试遵循this和this将模型添加到我的android项目中。一切如何产生相同的错误，即java.lang.IllegalArgumentException: ByteBuffer is not a valid flatbuffer model

用于构建模型的代码

model = Sequential()
model.add(LSTM(128, input_shape=(3000,5), return_sequences=True))
model.add(Dropout(0.2))
model.add(BatchNormalization())

model.add(LSTM(128, return_sequences=True))
model.add(Dropout(0.2))
model.add(BatchNormalization())

model.add(LSTM(128))
model.add(Dropout(0.1))
model.add(BatchNormalization())

model.add(Dense(64))
model.add(Dropout(0.2))

model.add(Dense(1, activation="sigmoid"))

opt = tf.keras.optimizers.Adam(lr=0.001, decay=1e-6)
model.compile(loss="binary_crossentropy",
             optimizer=opt,
             metrics=["accuracy"])


history = model.fit(X,
                   y,
                   batch_size=64,
                   epochs=10,
                   callbacks=[callbacks],
                   verbose=1)

完整的错误日志

E/ModelResourceManager: Error preloading model resource
    com.google.firebase.ml.common.FirebaseMLException: Local model load failed with the model options: Local model path: drowsy-detector-v21.tflite. Remote model name: unspecified. 
        at com.google.firebase.ml.common.internal.modeldownload.zzj.zza(com.google.firebase:firebase-ml-common@@22.1.0:36)
        at com.google.android.gms.internal.firebase_ml.zzrj.zza(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:111)
        at com.google.android.gms.internal.firebase_ml.zzrj.zzol(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:107)
        at com.google.android.gms.internal.firebase_ml.zzqr.zzf(com.google.firebase:firebase-ml-common@@22.1.0:53)
        at com.google.android.gms.internal.firebase_ml.zzqr$zza.zzoo(com.google.firebase:firebase-ml-common@@22.1.0:7)
        at com.google.android.gms.internal.firebase_ml.zzqr$zza.call(com.google.firebase:firebase-ml-common@@22.1.0:24)
        at com.google.android.gms.internal.firebase_ml.zzpx.zza(com.google.firebase:firebase-ml-common@@22.1.0:32)
        at com.google.android.gms.internal.firebase_ml.zzpw.run(Unknown Source:4)
        at android.os.Handler.handleCallback(Handler.java:873)
        at android.os.Handler.dispatchMessage(Handler.java:99)
        at com.google.android.gms.internal.firebase_ml.zze.dispatchMessage(com.google.firebase:firebase-ml-common@@22.1.0:6)
        at android.os.Looper.loop(Looper.java:201)
        at android.os.HandlerThread.run(HandlerThread.java:65)
     Caused by: java.lang.IllegalArgumentException: ByteBuffer is not a valid flatbuffer model
        at org.tensorflow.lite.NativeInterpreterWrapper.createModelWithBuffer(Native Method)
        at org.tensorflow.lite.NativeInterpreterWrapper.<init>(NativeInterpreterWrapper.java:59)
        at org.tensorflow.lite.Interpreter.<init>(Interpreter.java:207)
        at com.google.android.gms.internal.firebase_ml.zzrj.zzb(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:174)
        at com.google.android.gms.internal.firebase_ml.zzrl.zzc(Unknown Source:0)
        at com.google.android.gms.internal.firebase_ml.zzrj.zza(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:170)
        at com.google.android.gms.internal.firebase_ml.zzrk.zza(Unknown Source:6)
        at com.google.firebase.ml.common.internal.modeldownload.zzj.zzb(com.google.firebase:firebase-ml-common@@22.1.0:61)
        at com.google.firebase.ml.common.internal.modeldownload.zzj.zza(com.google.firebase:firebase-ml-common@@22.1.0:21)
        at com.google.android.gms.internal.firebase_ml.zzrj.zza(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:111) 
        at com.google.android.gms.internal.firebase_ml.zzrj.zzol(com.google.firebase:firebase-ml-model-interpreter@@22.0.2:107) 
        at com.google.android.gms.internal.firebase_ml.zzqr.zzf(com.google.firebase:firebase-ml-common@@22.1.0:53) 
        at com.google.android.gms.internal.firebase_ml.zzqr$zza.zzoo(com.google.firebase:firebase-ml-common@@22.1.0:7) 
        at com.google.android.gms.internal.firebase_ml.zzqr$zza.call(com.google.firebase:firebase-ml-common@@22.1.0:24) 
        at com.google.android.gms.internal.firebase_ml.zzpx.zza(com.google.firebase:firebase-ml-common@@22.1.0:32) 
        at com.google.android.gms.internal.firebase_ml.zzpw.run(Unknown Source:4) 
        at android.os.Handler.handleCallback(Handler.java:873) 
        at android.os.Handler.dispatchMessage(Handler.java:99) 
        at com.google.android.gms.internal.firebase_ml.zze.dispatchMessage(com.google.firebase:firebase-ml-common@@22.1.0:6) 
        at android.os.Looper.loop(Looper.java:201) 
        at android.os.HandlerThread.run(HandlerThread.java:65)

环境

OS平台：Google Colab，Android 9（Poco F1）
TensorFlow版本：2.2.0-rc3（也通过pip每晚安装一次）
firebase ml模型解释器版本：firebase-ml-model-interpreter：22.0.2

我非常感激这个项目的任何想法现在都非常重要。谢谢

LSTM模型转换为TFLite可以正常工作，但导入到Android项目时出错

0 个答案: