Python人臉識別微笑檢測

import cv2                     #   影像處理的庫 OpenCv

import dlib                    # 人臉識別的庫 dlib

import numpy as np             # 資料處理的庫 numpy

class face_emotion():

     def __init__(self):

         self.detector = dlib.get_frontal_face_detector()

         self.predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")

         self.cap = cv2.VideoCapture(0)

         self.cap.set(3, 480)

         self.cnt = 0  

     def learning_face(self):

         line_brow_x = []

         line_brow_y = []

         while(self.cap.isOpened()):

             flag, im_rd = self.cap.read()

             k = cv2.waitKey(1)

             # 取灰度

             img_gray = cv2.cvtColor(im_rd, cv2.COLOR_RGB2GRAY)  

             faces = self.detector(img_gray, 0)

             font = cv2.FONT_HERSHEY_SIMPLEX

             # 如果檢測到人臉

             if(len(faces) != 0):

                 # 對每個人臉都標出 68 個特徵點

                 for i in range(len(faces)):

                     for k, d in enumerate(faces):

                         cv2.rectangle(im_rd, (d.left(), d.top()), (d.right(), d.bottom()), (0,0,255))

                         self.face_width = d.right() - d.left()

                         shape = self.predictor(im_rd, d)

                         mouth_width = (shape.part(54).x - shape.part(48).x) / self.face_width

                         mouth_height = (shape.part(66).y - shape.part(62).y) / self.face_width

                         brow_sum = 0

                         frown_sum = 0

                         for j in range(17, 21):

                             brow_sum += (shape.part(j).y - d.top()) + (shape.part(j + 5).y - d.top())

                             frown_sum += shape.part(j + 5).x - shape.part(j).x

                             line_brow_x.append(shape.part(j).x)

                             line_brow_y.append(shape.part(j).y)

                         tempx = np.array(line_brow_x)

                         tempy = np.array(line_brow_y)

                         z1 = np.polyfit(tempx, tempy, 1)  

                         self.brow_k = -round(z1[0], 3)

                         brow_height = (brow_sum / 10) / self.face_width # 眉毛高度佔比

                         brow_width = (frown_sum / 5) / self.face_width  # 眉毛距離佔比

                         eye_sum = 外匯跟單gendan5.com(shape.part(41).y - shape.part(37).y + shape.part(40).y - shape.part(38).y +

                                    shape.part(47).y - shape.part(43).y + shape.part(46).y - shape.part(44).y)

                         eye_hight = (eye_sum / 4) / self.face_width

                         if round(mouth_height >= 0.03) and eye_hight<0.56:

                             cv2.putText(im_rd, "smile", (d.left(), d.bottom() + 20), cv2.FONT_HERSHEY_SIMPLEX, 2,

                                             (0,255,0), 2, 4)

                         if round(mouth_height<0.03) and self.brow_k>-0.3:

                             cv2.putText(im_rd, "unsmile", (d.left(), d.bottom() + 20), cv2.FONT_HERSHEY_SIMPLEX, 2,

                                         (0,255,0), 2, 4)

                 cv2.putText(im_rd, "Face-" + str(len(faces)), (20,50), font, 0.6, (0,0,255), 1, cv2.LINE_AA)

             else:

                 cv2.putText(im_rd, "No Face", (20,50), font, 0.6, (0,0,255), 1, cv2.LINE_AA)

             im_rd = cv2.putText(im_rd, "S: screenshot", (20,450), font, 0.6, (255,0,255), 1, cv2.LINE_AA)

             im_rd = cv2.putText(im_rd, "Q: quit", (20,470), font, 0.6, (255,0,255), 1, cv2.LINE_AA)

             if (cv2.waitKey(1) & 0xFF) == ord('s'):

                 self.cnt += 1

                 cv2.imwrite("screenshoot" + str(self.cnt) + ".jpg", im_rd)

             # 按下 q 鍵退出

             if (cv2.waitKey(1)) == ord('q'):

                 break

             # 視窗顯示

             cv2.imshow("Face Recognition", im_rd)

         self.cap.release()

         cv2.destroyAllWindows()

if __name__ == "__main__":

     my_face = face_emotion()

     my_face.learning_face()