def run(self):
state = ['s', '1', '5', '8']
ind = random.randint(0, 11)
self.order_value.emit(str(state[ind % 4]))
if str(state[ind % 4]) == 's':
speech.say("请保持沉默。")
else:
speech.say("请说" + str(state[ind % 4]))
print('请说 ' + str(state[ind % 4]))
while True:
m = 0
frames = []
speech.say(3)
speech.say(2)
speech.say(1)
while m < 3:
ret, img = self.cap.read()
cv2.waitKey(33)
if ret: # 如果摄像头读取图像成功
frame = np.array(img)
frames.append(frame)
# cv2.waitKey(60)
m += 1
mouth_points = get_frames_mouth(self.detector, self.predictor, frames)
print(np.shape(mouth_points))
if np.shape(mouth_points) != (np.shape(mouth_points)[0], 2, 12):
continue
mouth_points_mean = np.mean(mouth_points, axis=0)
mouth_points_mean_re = np.reshape(mouth_points_mean, newshape=[1, 24])
pred = self.svmmodel.predict(mouth_points_mean_re)
self.pre_value.emit(pred[0])
print('我认为你说了 ' + pred[0])
if pred == state[ind % 4]:
true = True
else:
true = False
time.sleep(2)
self.order_value.emit('')
self.pre_value.emit('')
if true:
print('正确!\n')
speech.say("正确。")
ind = random.randint(0, 11)
self.order_value.emit(str(state[ind % 4]))
if str(state[ind % 4]) == 's':
speech.say("请保持沉默。")
else:
speech.say("请说" + str(state[ind % 4]))
print('请说 ' + str(state[ind % 4]))
else:
speech.say("我认为你说错了，请继续说" + str(state[ind % 4]))
print('错误！请继续说 ' + str(state[ind % 4]))
self.order_value.emit(str(state[ind % 4]))
def run(self):
cap = cv2.VideoCapture(0, cv2.CAP_DSHOW)
FACE_PREDICTOR_PATH = r'D:\LipNum\shape_predictor_68_face_landmarks.dat'
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor(FACE_PREDICTOR_PATH)
modelpath = os.path.join(os.path.abspath('..'), 'model', 'svm_model.m')
self.svmmodel = joblib.load(modelpath)
true = True
state = ['s', '1', '5', '8']
ind = random.randint(0, 11)
self.order_value.emit(str(state[ind % 4]))
print('请说 ' + str(state[ind % 4]))
while cap.isOpened(): # isOpened() 检测摄像头是否处于打开状态
m = 0
frames = []
if true:
print('正确!\n')
ind = random.randint(0, 11)
self.order_value.emit(str(state[ind % 4]))
print('请说 ' + str(state[ind % 4]))
else:
print('错误！请继续说 ' + str(state[ind % 4]))
self.order_value.emit(str(state[ind % 4]))
# if k == 107: # ASCLL(k)==127 keep
time.sleep(1)
while m < 10:
ret, img = cap.read() # 把摄像头获取的图像信息保存之img变量
if ret: # 如果摄像头读取图像成功
cv2.imwrite(r'image\image{}.jpg'.format(m), img)
frame = np.array(img)
frames.append(frame)
cv2.waitKey(60)
m += 1
mouth_points = get_frames_mouth(detector, predictor, frames)
mouth_points_mean = np.mean(mouth_points, axis=0)
# print(mouth_points_mean.shape)
mouth_points_mean_re = np.reshape(mouth_points_mean,
newshape=[1, 24])
# print(mouth_points_mean_re.shape)
# mouth_points_mean_re.reshape(-1, 1)
# print(mouth_points_mean_re.shape)
pred = self.svmmodel.predict(mouth_points_mean_re)
self.pre_value.emit(pred[0])
print('我认为你说了 ' + pred[0])
if pred == state[ind % 4]:
true = True
else:
true = False
time.sleep(2)
self.order_value.emit('')
self.pre_value.emit('')
def run(self):
while self.cap.isOpened(): # isOpened() 检测摄像头是否处于打开状态
m = 0
frames = []
while m < 2:
ret, img = self.cap.read() # 把摄像头获取的图像信息保存之img变量
cv2.waitKey(33)
if ret: # 如果摄像头读取图像成功
frame = np.array(img)
frames.append(frame)
m += 1
mouth_points = get_frames_mouth(self.detector, self.predictor, frames)
print(np.shape(mouth_points))
if np.shape(mouth_points) != (np.shape(mouth_points)[0], 2, 12):
continue
mouth_points_mean = np.mean(mouth_points, axis=0)
mouth_points_mean_re = np.reshape(mouth_points_mean, newshape=[1, 24])
pred = self.svmmodel.predict(mouth_points_mean_re)
self.predict.emit(pred[0])
print('我认为你现在说了 ' + pred[0])
ret, img = cap.read() # 把摄像头获取的图像信息保存之img变量
if ret: # 如果摄像头读取图像成功
print(np.shape(img))
cv2.imshow('Image', img)
while m < 10:
ret, img = cap.read() # 把摄像头获取的图像信息保存之img变量
if ret: # 如果摄像头读取图像成功
cv2.imshow('Image', img)
cv2.imwrite(r'image\image{}.jpg'.format(m), img)
frame = np.array(img)
frames.append(frame)
k = cv2.waitKey(5)
m += 1
mouth_points = get_frames_mouth(detector, predictor, frames)
if np.shape(mouth_points) != (np.shape(mouth_points)[0], 2, 12):
continue
mouth_points_mean = np.mean(mouth_points, axis=0)
# print(mouth_points_mean.shape)
mouth_points_mean_re = np.reshape(mouth_points_mean, newshape=[1, 24])
# print(mouth_points_mean_re.shape)
# mouth_points_mean_re.reshape(-1, 1)
# print(mouth_points_mean_re.shape)
pred = svmmodel.predict(mouth_points_mean_re)
print('我认为你现在说了 ' + pred)
# plt.figure(1)
# for m in range(np.shape(X_re)[0]):
# if Y[m] == '1':
