model.add(Dropout(0.2))
model.add(Conv2D(filters=128 * 2, kernel_size=2, activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.2))
model.add(GlobalAveragePooling2D())
model.add(Dense(num_labels, activation='softmax'))
#----------------------------------------------------------------------------------------------#
# Load pretrained softmax-crossEntropyLoss model
if False:
model.load_weights(
filepath='./saved_models_Keras/weights.best.basic_cnn.hdf5')
print('Loaded Model from disk')
# Set every layer to be non-trainable:
for k, v in model._get_trainable_state().items():
k.trainable = False
#----------------------------------------------------------------------------------------------#
# Print current trainable map
#print(model._get_trainable_state())
#----------------------------------------------------------------------------------------------#
# Plot the model
# Need to install pydot and graphviz
# plot_model(model, show_shapes=True, show_layer_names = True)
#----------------------------------------------------------------------------------------------#
# Compile the model
if False:
opt = optimizers.Adam(lr=0.001,
beta_1=0.9,
beta_2=0.999,
epsilon=1e-7,
model_pretrained.add(Conv2D(filters=128 * 2, kernel_size=2, activation='relu'))
model_pretrained.add(MaxPooling2D(pool_size=2))
model_pretrained.add(Dropout(0.2))
model_pretrained.add(GlobalAveragePooling2D())
model_pretrained.add(Dense(num_labels, activation='softmax'))
#----------------------------------------------------------------------------------------------#
# Load pretrained softmax-crossEntropyLoss model
model_pretrained.load_weights(filepath='weights.best.basic_cnn.hdf5')
print('Loaded Model from disk')
# Print current trainable map
#print(model_pretrained._get_trainable_state())
# Set every layer to be non-trainable:
for k, v in model_pretrained._get_trainable_state().items():
k.trainable = False
#----------------------------------------------------------------------------------------------#
# Construct model
model = Sequential()
model.add(model_pretrained)
#----------------------------------------------------------------------------------------------#
# --> Plot the model
# --> Need to install pydot and graphviz
#plot_model(model, show_shapes=True, show_layer_names = True)
#----------------------------------------------------------------------------------------------#
# --> Predict individual sounds.
def print_prediction(file_name, label):
ground_label = {
