利用cnn卷积神经网络搭建了一个识别是否戴口罩的模型,并且通过opencv对画面进行实时检测。
import cv2
import tensorflow as tf
import numpy as np
from scipy.special import softmax
import dlib
class mask():
def __init__(self):
self.model = tf.keras.models.load_model(r'D:\PYTHON\shizhanxiangmu\CNN\cnn人脸口罩佩戴识别\mark_model')
self.face_mode = dlib.get_frontal_face_detector()
def face_det(self,img):
face_id = self.face_mode(img,1)
for face in face_id:
l, t, r, b = face.left(), face.top(), face.right(), face.bottom()
cv2.rectangle(img,(l,t),(r,b),(200,0,190),5)
return img
def blob_pre(self,img):
#人脸检测器
img = self.face_det(img)
# 转为blob
img_blob = cv2.dnn.blobFromImage(img, 1, (100, 100), (104, 177, 123), swapRB=True)
# 压缩维度、转置
img_squeeze = np.squeeze(img_blob).T
# 旋转
img_a = cv2.rotate(img_squeeze, cv2.ROTATE_90_CLOCKWISE)
# 镜像翻转
img = cv2.flip(img_a, 1)
# 把负数变为0
img_blob = np.maximum(img, 0) / img.max()
#变成4维
img_blob = img_blob.reshape(1,100,100,3)
#模型预测
result = self.model.predict(img_blob)
results = softmax(result[0])
max_index = results.argmax() #最大值索引
max_val = results[max_index] #索引最大值
line =['mask','no_mask']
txt = f'{line[max_index]}:{round(max_val*100,2)}'
return txt
def rec(self):
cap = cv2.VideoCapture(0)
while True:
ret ,frame = cap.read()
frame = cv2.flip(frame,1)
txt= self.blob_pre(frame)
cv2.putText(frame,txt, (10, 50), cv2.FONT_HERSHEY_COMPLEX, 1, (0, 255, 0), 1)
cv2.imshow('img',frame)
if ord('q') == cv2.waitKey(1):
break
cap.release()
cv2.destroyAllWindows()
s=mask()
s.rec()