如何使用Python保存和加载自定义构建的MobileNetV2模型

Question

我正在尝试保存一个自定义的构建模型，但遇到了像这样的错误：

Error During Save: ValueError: Unable to synchronously create dataset (name already exists)
Error During Load: KeyError: "Unable to synchronously open object (object 'time_distributed_2' doesn't exist)"**

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
Cell In[36], line 2
      1 # Load the model with custom objects
----> 2 loaded_model = load_model("MoBiLSTM_model.h5", custom_objects={"MobileNetV2": MobileNetV2, "time_distributed_2":TimeDistributed})

File ~/akash_ai_ml/video_analysis/.venv/lib64/python3.11/site-packages/keras/src/saving/saving_api.py:183, in load_model(filepath, custom_objects, compile, safe_mode)
    176     return saving_lib.load_model(
    177         filepath,
    178         custom_objects=custom_objects,
    179         compile=compile,
    180         safe_mode=safe_mode,
    181     )
    182 if str(filepath).endswith((".h5", ".hdf5")):
--> 183     return legacy_h5_format.load_model_from_hdf5(filepath)
    184 elif str(filepath).endswith(".keras"):
    185     raise ValueError(
    186         f"File not found: filepath={filepath}. "
    187         "Please ensure the file is an accessible `.keras` "
    188         "zip file."
    189     )

File ~/akash_ai_ml/video_analysis/.venv/lib64/python3.11/site-packages/keras/src/legacy/saving/legacy_h5_format.py:138, in load_model_from_hdf5(filepath, custom_objects, compile)
    133     model = saving_utils.model_from_config(
    134         model_config, custom_objects=custom_objects
...
File h5py/_objects.pyx:55, in h5py._objects.with_phil.wrapper()

File h5py/h5o.pyx:189, in h5py.h5o.open()

KeyError: "Unable to synchronously open object (object 'time_distributed_2' doesn't exist)"

这是我的代码：

# Specify the height and width to which each video frame will be resized in our dataset.
IMAGE_HEIGHT , IMAGE_WIDTH = 64, 64

# Specify the number of frames of a video that will be fed to the model as one sequence.
SEQUENCE_LENGTH = 2


DATASET_DIR = "voilence_dataset/Real Life Violence Dataset"

CLASSES_LIST = ["NonViolence","Violence"]

from keras.applications.mobilenet_v2 import MobileNetV2

mobilenet = MobileNetV2( include_top=False , weights="imagenet")

#Fine-Tuning to make the last 40 layer trainable
mobilenet.trainable=True

for layer in mobilenet.layers[:-40]:
  layer.trainable=False

#mobilenet.summary()

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Input, TimeDistributed, Dropout, Flatten, 
Bidirectional, LSTM, Dense
from keras.applications.mobilenet_v2 import MobileNetV2

def create_model(input_shape, num_classes, mobilenet):
    model = Sequential()

    model.add(TimeDistributed(mobilenet, input_shape=input_shape))
    model.add(Dropout(0.25))
    model.add(TimeDistributed(Flatten()))

    lstm_fw = LSTM(units=32)
    lstm_bw = LSTM(units=32, go_backwards=True)
    model.add(Bidirectional(lstm_fw, backward_layer=lstm_bw))

    model.add(Dropout(0.25))
    model.add(Dense(256, activation='relu'))
    model.add(Dropout(0.25))
    model.add(Dense(128, activation='relu'))
    model.add(Dropout(0.25))
    model.add(Dense(64, activation='relu'))
    model.add(Dropout(0.25))
    model.add(Dense(32, activation='relu'))
    model.add(Dropout(0.25))
    model.add(Dense(num_classes, activation='softmax'))

    model.summary()

    return model


input_shape = (SEQUENCE_LENGTH, IMAGE_HEIGHT, IMAGE_WIDTH, 3)
num_classes = len(CLASSES_LIST)
MoBiLSTM_model = create_model(input_shape, num_classes, mobilenet)

# Construct the model
MoBiLSTM_model = create_model(input_shape, num_classes, mobilenet)

# Plot the structure of the constructed model
plot_model(MoBiLSTM_model, to_file='MobBiLSTM_model_structure_plot.png', 
show_shapes=True, show_layer_names=True)

# Create Early Stopping Callback to monitor the accuracy
early_stopping_callback = EarlyStopping(monitor = 'val_accuracy', patience = 10, 
restore_best_weights = True)

# Create ReduceLROnPlateau Callback to reduce overfitting by decreasing learning
reduce_lr = tf.keras.callbacks.ReduceLROnPlateau(monitor='val_loss',
                                              factor=0.6,
                                              patience=5,
                                              min_lr=0.00005,
                                              verbose=1)

# Compiling the model 
MoBiLSTM_model.compile(loss = 'categorical_crossentropy', optimizer = 'sgd', metrics = 
["accuracy"])

# Fitting the model 
MobBiLSTM_model_history = MoBiLSTM_model.fit(x = features_train, y = labels_train, 
epochs = 30, batch_size = 8 ,
                                         shuffle = True, validation_split = 0.2, 
callbacks = [early_stopping_callback,reduce_lr])

# Define the file path where you want to save the model
model_save_path = 'model_new.h5'

# Save the trained model
MoBiLSTM_model.save(model_save_path)

print("Model saved successfully at:", model_save_path)

深度学习模型保存 MobileNetV2

如何使用Python保存和加载自定义构建的MobileNetV2模型

1 个回答

撰写回答