def predict(self, video_file_path):
images, preprocessed_images, x = extract_vgg16_features_live(self.vgg16_model, video_file_path)
frames = x.shape[0]
if frames > self.expected_frames:
x = x[0:self.expected_frames, :]
images = images[0:self.expected_frames, :]
preprocessed_images = preprocessed_images[0:self.expected_frames, :]
elif frames < self.expected_frames:
temp = np.zeros(shape=(self.expected_frames, x.shape[1]))
temp[0:frames, :] = x
x = temp
shape1 = list(images.shape)
shape1[0] = self.expected_frames
shape1 = tuple(shape1)
temp1 = np.zeros(shape=shape1)
temp1[0:frames, :] = images
images = temp1
shape2 = list(preprocessed_images.shape)
shape2[0] = self.expected_frames
shape2 = tuple(shape2)
temp2 = np.zeros(shape=shape2)
temp2[0:frames, :] = preprocessed_images
preprocessed_images = temp2
last_activation = self.model.predict(np.array([x]))
predicted_class = np.argmax(last_activation[0])
predicted_label = self.labels_idx2word[predicted_class]
return images, preprocessed_images, x, predicted_label, np.max(last_activation[0]), last_activation[0]
def predict(self, video_file_path):
images, preprocessed_images, x = extract_vgg16_features_live(self.vgg16_model, video_file_path)
frames = x.shape[0]
if frames > self.expected_frames:
x = x[0:self.expected_frames, :]
elif frames < self.expected_frames:
temp = np.zeros(shape=(self.expected_frames, x.shape[1]))
temp[0:frames, :] = x
x = temp
predicted_class = np.argmax(self.model.predict(np.array([x]))[0])
predicted_label = self.labels_idx2word[predicted_class]
return predicted_label