def predict(weight_path, video_path, absolute_max_string_len=32, output_size=28):
print ("\nLoading data from disk...")
video = Video(vtype='face', face_predictor_path=FACE_PREDICTOR_PATH)
if os.path.isfile(video_path):
video.from_video(video_path)
else:
video.from_frames(video_path)
print ("Data loaded.\n")
if K.image_data_format() == 'channels_first':
img_c, frames_n, img_w, img_h = video.data.shape
else:
frames_n, img_w, img_h, img_c = video.data.shape
lipnet = LipNet(img_c=img_c, img_w=img_w, img_h=img_h, frames_n=frames_n,
absolute_max_string_len=absolute_max_string_len, output_size=output_size)
adam = Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08)
lipnet.model.compile(loss={'ctc': lambda y_true, y_pred: y_pred}, optimizer=adam)
lipnet.model.load_weights(weight_path)
spell = Spell(path=PREDICT_DICTIONARY)
decoder = Decoder(greedy=PREDICT_GREEDY, beam_width=PREDICT_BEAM_WIDTH,
postprocessors=[labels_to_text, spell.sentence])
X_data = np.array([video.data]).astype(np.float32) / 255
input_length = np.array([len(video.data)])
y_pred = lipnet.predict(X_data)
result = decoder.decode(y_pred, input_length)[0]
return (video, result)
def predict(weight_path, video):
global lipnet
global adam
global spell
global decoder
if lipnet is None:
lipnet = LipNet(img_c=3,
img_w=100,
img_h=50,
frames_n=75,
absolute_max_string_len=32,
output_size=28)
adam = Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08)
lipnet.model.compile(loss={
'ctc': lambda y_true, y_pred: y_pred
},
optimizer=adam)
lipnet.model.load_weights(weight_path)
spell = Spell(path=PREDICT_DICTIONARY)
decoder = Decoder(greedy=PREDICT_GREEDY,
beam_width=PREDICT_BEAM_WIDTH,
postprocessors=[labels_to_text, spell.sentence])
X_data = np.array([video.data]).astype(np.float32) / 255
input_length = np.array([len(video.data)])
y_pred = lipnet.predict(X_data)
result = decoder.decode(y_pred, input_length)[0]
show_video_subtitle(video.face, result)
print result
def predict_videos(video_data,
weight_path,
absolute_max_string_len=32,
output_size=28):
if K.image_data_format() == 'channels_first':
img_c, frames_n, img_w, img_h = video_data.shape
else:
frames_n, img_w, img_h, img_c = video_data.shape
lipnet = LipNet(img_c=img_c,
img_w=img_w,
img_h=img_h,
frames_n=frames_n,
absolute_max_string_len=absolute_max_string_len,
output_size=output_size)
adam = Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=1e-08)
lipnet.model.compile(loss={
'ctc': lambda y_true, y_pred: y_pred
},
optimizer=adam)
lipnet.model.load_weights(weight_path)
spell = Spell(path=PREDICT_DICTIONARY)
decoder = Decoder(greedy=PREDICT_GREEDY,
beam_width=PREDICT_BEAM_WIDTH,
postprocessors=[labels_to_text, spell.sentence])
X_data = np.array([video_data]).astype(np.float32) / 255
input_length = np.array([len(video_data)])
y_pred = lipnet.predict(X_data)
#print(y_pred[0,0])
#print(y_pred[0,40])
result = decoder.decode(y_pred, input_length)[0]
return result