depth_multiplier=1,
dropout=0.001,
include_top=False,
weights="imagenet",
input_shape=(224, 224,
3))
model = GlobalAveragePooling2D()(base_model.output)
model = Dropout(0.001)(model)
output_layer = Dense(n_classes, activation='softmax')(model)
model = Model(base_model.input, output_layer)
# Load saved weights
model.load_weights(savepath + 'weights.h5', by_name=True)
# Make predictions
preds = model.predict(prediction_generator)
preds_class_indices = preds.argmax(axis=-1)
# Convert labels and predictions to dictionaries
labels = dict((v, k) for k, v in classes.items())
predictions = [labels[k] for k in preds_class_indices]
# Put them together with the filenames into a dataframe
filenames = prediction_generator.filenames
results = pd.DataFrame({"Filename": filenames, "Prediction": predictions})
print('')
print(results)
print('')
print('')
output_layer = Dense(6, activation='softmax')(model)
model = Model(base_model.input, output_layer)
# Load saved weights
model.load_weights(savepath + 'weights.h5', by_name=True)
# Load and preprocess image
img = image.load_img(imagepath, target_size=(224, 224))
z = image.img_to_array(img)
z = np.expand_dims(z, axis=0)
z = preprocess_input(z)
# Make prediction
preds = model.predict(z)
maximum_model_output = model.output[:, 0]
last_conv_layer = model.layers[83]
# Pooled grads of last convolutional layer and iterate over image
grads = K.gradients(model.output[:, 0], last_conv_layer.output)[0]
pooled_grads = K.mean(grads, axis=(0, 1, 2))
iterate = K.function([model.input],
[pooled_grads, last_conv_layer.output[0]])
pooled_grads_value, conv_layer_output_value = iterate([z])
# Extract 768 pooled grads of last convolutional layer
for i in range(768):
conv_layer_output_value[:, :, i] *= pooled_grads_value[i]