face_encodings method can not run on asynchronous threads

• face_recognition version:
• Python version: 3.4
• Operating System: windows 7

Description

I want to use face_recognition.face_encodings method on asynchronous threads, Because it causes my webcam to be unsmooth, but when I use it on asynchronous threads, It has no effect at all

class CompareByThreadPool(object):
    executor = ThreadPoolExecutor(max_workers=1)
   
    def __init__(self):

        self.mutex = QMutex()
        self.obama_img = []
        self.obama_face_encoding = []

        self.face_locations = []
        self.face_encodings = []
        self.face_names = []
        self.process_this_frame = False
        self.can_begain_compare = False
        self.can_begain_encoding = False
        self.frame=numpy.arange(0)
        future = self.executor.submit(self.cricleRun,self.frame)

    def set_frame(self,frame):
        self.frame = frame

    def cricleRun(self,frame):
        while True:
            print('self.frame.size: ',self.frame.size)
            if self.frame.size > 0:
                self.face_rec(self.frame)
            time.sleep(0.4)
    
    def face_rec(self,frame):
        
        small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25)
    
        # Convert the image from BGR color (which OpenCV uses) to RGB color (which face_recognition uses)
        rgb_small_frame = small_frame[:, :, ::-1]
        # rgb_small_frame = small_frame
      
        self.face_locations = face_recognition.face_locations(rgb_small_frame)

        rec_begain_time = time.time()
        print('begain time: ',rec_begain_time)
        if self.process_this_frame:
            self.face_encodings = face_recognition.face_encodings(rgb_small_frame, self.face_locations)
        self.process_this_frame = not self.process_this_frame
        print('face_rec compareFlag: ',self.can_begain_compare)
        if self.can_begain_compare:
            face_names = []
            for face_encoding in self.face_encodings:
                #match = face_recognition.compare_faces([obama_face_encoding], face_encoding,0.5)
                face_distances = face_recognition.face_distance([self.obama_face_encoding],face_encoding);
                print("face_distance: ",face_distances[0]);
                #if match[0]:
                if face_distances[0] < 0.6:
                    name = "Success"
                else:
                    name = "unknown"
                rec_end_time= time.time()
                face_names.append(name)
                print("name : {0} time: {1} ".format(name,rec_end_time-rec_begain_time))

What I Did

Can it run on asynchronous threads ， Please tell me how can I make my webcam smooth

Start with the face_recognition.face_encodings call and the ThreadPoolExecutor-based CompareByThreadPool example in the issue, then examine how frames and shared attributes are passed through cricleRun and face_rec. Reproduce the behavior on the stated Python 3.4 and Windows 7 setup; done would require a confirmed explanation or documented change that makes asynchronous processing affect webcam smoothness.