Python怎么实现人脸识别微笑检测
短信预约 -IT技能 免费直播动态提醒
这篇文章主要介绍“Python怎么实现人脸识别微笑检测”，在日常操作中，相信很多人在Python怎么实现人脸识别微笑检测问题上存在疑惑，小编查阅了各式资料，整理出简单好用的操作方法，希望对大家解答”Python怎么实现人脸识别微笑检测”的疑惑有所帮助！接下来，请跟着小编一起来学习吧！
一.实验准备
环境搭建
pip install tensorflow==1.2.0pip install keras==2.0.6pip install dlib==19.6.1pip install h6py==2.10如果是新建虚拟环境，还需安装以下包
pip install opencv_python==4.1.2.30pip install pillowpip install matplotlibpip install h6py使用genki-4k数据集
可从此处下载
二.图片预处理
打开数据集
我们需要将人脸检测出来并对图片进行裁剪
代码如下：
import dlib # 人脸识别的库dlibimport numpy as np # 数据处理的库numpyimport cv2 # 图像处理的库OpenCvimport os # dlib预测器detector = dlib.get_frontal_face_detector()predictor = dlib.shape_predictor('D:\\shape_predictor_68_face_landmarks.dat') # 读取图像的路径path_read = "C:\\Users\\28205\\Documents\\Tencent Files\\2820535964\\FileRecv\\genki4k\\files"num=0for file_name in os.listdir(path_read):#aa是图片的全路径 aa=(path_read +"/"+file_name) #读入的图片的路径中含非英文 img=cv2.imdecode(np.fromfile(aa, dtype=np.uint8), cv2.IMREAD_UNCHANGED) #获取图片的宽高 img_shape=img.shape img_height=img_shape[0] img_width=img_shape[1] # 用来存储生成的单张人脸的路径 path_save="C:\\Users\\28205\\Documents\\Tencent Files\\2820535964\\FileRecv\\genki4k\\files1" # dlib检测 dets = detector(img,1) print("人脸数：", len(dets)) for k, d in enumerate(dets): if len(dets)>1: continue num=num+1 # 计算矩形大小 # (x,y), (宽度width, 高度height) pos_start = tuple([d.left(), d.top()]) pos_end = tuple([d.right(), d.bottom()]) # 计算矩形框大小 height = d.bottom()-d.top() width = d.right()-d.left() # 根据人脸大小生成空的图像 img_blank = np.zeros((height, width, 3), np.uint8) for i in range(height): if d.top()+i>=img_height:# 防止越界 continue for j in range(width): if d.left()+j>=img_width:# 防止越界 continue img_blank[i][j] = img[d.top()+i][d.left()+j] img_blank = cv2.resize(img_blank, (200, 200), interpolation=cv2.INTER_CUBIC) cv2.imencode('.jpg', img_blank)[1].tofile(path_save+"\\"+"file"+str(num)+".jpg") # 正确方法运行效果如下：
共识别出3878张图片。
某些图片没有识别出人脸，所以没有裁剪保存，可以自行添加图片补充。
三.划分数据集
代码：
import os, shutil# 原始数据集路径original_dataset_dir = 'C:\\Users\\28205\\Documents\\Tencent Files\\2820535964\\FileRecv\\genki4k\\files1'# 新的数据集base_dir = 'C:\\Users\\28205\\Documents\\Tencent Files\\2820535964\\FileRecv\\genki4k\\files2'os.mkdir(base_dir)# 训练图像、验证图像、测试图像的目录train_dir = os.path.join(base_dir, 'train')os.mkdir(train_dir)validation_dir = os.path.join(base_dir, 'validation')os.mkdir(validation_dir)test_dir = os.path.join(base_dir, 'test')os.mkdir(test_dir)train_cats_dir = os.path.join(train_dir, 'smile')os.mkdir(train_cats_dir)train_dogs_dir = os.path.join(train_dir, 'unsmile')os.mkdir(train_dogs_dir)validation_cats_dir = os.path.join(validation_dir, 'smile')os.mkdir(validation_cats_dir)validation_dogs_dir = os.path.join(validation_dir, 'unsmile')os.mkdir(validation_dogs_dir)test_cats_dir = os.path.join(test_dir, 'smile')os.mkdir(test_cats_dir)test_dogs_dir = os.path.join(test_dir, 'unsmile')os.mkdir(test_dogs_dir)# 复制1000张笑脸图片到train_c_dirfnames = ['file{}.jpg'.format(i) for i in range(1,900)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(train_cats_dir, fname) shutil.copyfile(class="lazy" data-src, dst)fnames = ['file{}.jpg'.format(i) for i in range(900, 1350)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(validation_cats_dir, fname) shutil.copyfile(class="lazy" data-src, dst) # Copy next 500 cat images to test_cats_dirfnames = ['file{}.jpg'.format(i) for i in range(1350, 1800)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(test_cats_dir, fname) shutil.copyfile(class="lazy" data-src, dst) fnames = ['file{}.jpg'.format(i) for i in range(2127,3000)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(train_dogs_dir, fname) shutil.copyfile(class="lazy" data-src, dst) # Copy next 500 dog images to validation_dogs_dirfnames = ['file{}.jpg'.format(i) for i in range(3000,3878)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(validation_dogs_dir, fname) shutil.copyfile(class="lazy" data-src, dst) # Copy next 500 dog images to test_dogs_dirfnames = ['file{}.jpg'.format(i) for i in range(3000,3878)]for fname in fnames: class="lazy" data-src = os.path.join(original_dataset_dir, fname) dst = os.path.join(test_dogs_dir, fname) shutil.copyfile(class="lazy" data-src, dst)运行效果如下：
四.CNN提取人脸识别笑脸和非笑脸
1.创建模型
代码：
#创建模型from keras import layersfrom keras import modelsmodel = models.Sequential()model.add(layers.Conv2D(32, (3, 3), activation='relu',input_shape=(150, 150, 3)))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(64, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(128, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(128, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Flatten())model.add(layers.Dense(512, activation='relu'))model.add(layers.Dense(1, activation='sigmoid'))model.summary()#查看运行效果：
2.归一化处理
代码：
#归一化from keras import optimizersmodel.compile(loss='binary_crossentropy', optimizer=optimizers.RMSprop(lr=1e-4), metrics=['acc'])from keras.preprocessing.image import ImageDataGeneratortrain_datagen = ImageDataGenerator(rescale=1./255)validation_datagen=ImageDataGenerator(rescale=1./255)test_datagen = ImageDataGenerator(rescale=1./255)train_generator = train_datagen.flow_from_directory( # 目标文件目录 train_dir, #所有图片的size必须是150x150 target_size=(150, 150), batch_size=20, # Since we use binary_crossentropy loss, we need binary labels class_mode='binary')validation_generator = test_datagen.flow_from_directory( validation_dir, target_size=(150, 150), batch_size=20, class_mode='binary')test_generator = test_datagen.flow_from_directory(test_dir, target_size=(150, 150), batch_size=20, class_mode='binary')for data_batch, labels_batch in train_generator: print('data batch shape:', data_batch.shape) print('labels batch shape:', labels_batch) break#'smile': 0, 'unsmile': 13.数据增强
代码：
#数据增强datagen = ImageDataGenerator( rotation_range=40, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode='nearest')#数据增强后图片变化import matplotlib.pyplot as plt# This is module with image preprocessing utilitiesfrom keras.preprocessing import imagefnames = [os.path.join(train_smile_dir, fname) for fname in os.listdir(train_smile_dir)]img_path = fnames[3]img = image.load_img(img_path, target_size=(150, 150))x = image.img_to_array(img)x = x.reshape((1,) + x.shape)i = 0for batch in datagen.flow(x, batch_size=1): plt.figure(i) imgplot = plt.imshow(image.array_to_img(batch[0])) i += 1 if i % 4 == 0: breakplt.show()运行效果：
4.创建网络
代码：
#创建网络model = models.Sequential()model.add(layers.Conv2D(32, (3, 3), activation='relu',input_shape=(150, 150, 3)))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(64, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(128, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Conv2D(128, (3, 3), activation='relu'))model.add(layers.MaxPooling2D((2, 2)))model.add(layers.Flatten())model.add(layers.Dropout(0.5))model.add(layers.Dense(512, activation='relu'))model.add(layers.Dense(1, activation='sigmoid'))model.compile(loss='binary_crossentropy', optimizer=optimizers.RMSprop(lr=1e-4), metrics=['acc'])#归一化处理train_datagen = ImageDataGenerator( rescale=1./255, rotation_range=40, width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, zoom_range=0.2, horizontal_flip=True,)test_datagen = ImageDataGenerator(rescale=1./255)train_generator = train_datagen.flow_from_directory( # This is the target directory train_dir, # All images will be resized to 150x150 target_size=(150, 150), batch_size=32, # Since we use binary_crossentropy loss, we need binary labels class_mode='binary')validation_generator = test_datagen.flow_from_directory( validation_dir, target_size=(150, 150), batch_size=32, class_mode='binary')history = model.fit_generator( train_generator, steps_per_epoch=100, epochs=60, validation_data=validation_generator, validation_steps=50)model.save('smileAndUnsmile1.h6')#数据增强过后的训练集与验证集的精确度与损失度的图形acc = history.history['acc']val_acc = history.history['val_acc']loss = history.history['loss']val_loss = history.history['val_loss']epochs = range(len(acc))plt.plot(epochs, acc, 'bo', label='Training acc')plt.plot(epochs, val_acc, 'b', label='Validation acc')plt.title('Training and validation accuracy')plt.legend()plt.figure()plt.plot(epochs, loss, 'bo', label='Training loss')plt.plot(epochs, val_loss, 'b', label='Validation loss')plt.title('Training and validation loss')plt.legend()plt.show()运行结果：
速度较慢，要等很久
5.单张图片测试
代码：
# 单张图片进行判断 是笑脸还是非笑脸import cv2from keras.preprocessing import imagefrom keras.models import load_modelimport numpy as np#加载模型model = load_model('smileAndUnsmile1.h6')#本地图片路径img_path='test.jpg'img = image.load_img(img_path, target_size=(150, 150))img_tensor = image.img_to_array(img)/255.0img_tensor = np.expand_dims(img_tensor, axis=0)prediction =model.predict(img_tensor) print(prediction)if prediction[0][0]>0.5: result='非笑脸'else: result='笑脸'print(result)
运行结果：
6.摄像头实时测试
代码：
#检测视频或者摄像头中的人脸import cv2from keras.preprocessing import imagefrom keras.models import load_modelimport numpy as npimport dlibfrom PIL import Imagemodel = load_model('smileAndUnsmile1.h6')detector = dlib.get_frontal_face_detector()video=cv2.VideoCapture(0)font = cv2.FONT_HERSHEY_SIMPLEXdef rec(img): gray=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY) dets=detector(gray,1) if dets is not None: for face in dets: left=face.left() top=face.top() right=face.right() bottom=face.bottom() cv2.rectangle(img,(left,top),(right,bottom),(0,255,0),2) img1=cv2.resize(img[top:bottom,left:right],dsize=(150,150)) img1=cv2.cvtColor(img1,cv2.COLOR_BGR2RGB) img1 = np.array(img1)/255. img_tensor = img1.reshape(-1,150,150,3) prediction =model.predict(img_tensor) if prediction[0][0]>0.5: result='unsmile' else: result='smile' cv2.putText(img, result, (left,top), font, 2, (0, 255, 0), 2, cv2.LINE_AA) cv2.imshow('Video', img)while video.isOpened(): res, img_rd = video.read() if not res: break rec(img_rd) if cv2.waitKey(1) & 0xFF == ord('q'): breakvideo.release()cv2.destroyAllWindows()运行结果：
五.Dlib提取人脸特征识别笑脸和非笑脸
代码：
import cv2 # 图像处理的库 OpenCvimport dlib # 人脸识别的库 dlibimport numpy as np # 数据处理的库 numpyclass 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 # 如果检测到人脸 if(len(faces) != 0): # 对每个人脸都标出68个特征点 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 # 眉毛高度占比 brow_width = (frown_sum / 5) / self.face_width # 眉毛距离占比 eye_sum = (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 键退出 if (cv2.waitKey(1)) == ord('q'): break # 窗口显示 cv2.imshow("Face Recognition", im_rd) self.cap.release() cv2.destroyAllWindows()if __name__ == "__main__": my_face = face_emotion() my_face.learning_face()运行结果：
到此，关于“Python怎么实现人脸识别微笑检测”的学习就结束了，希望能够解决大家的疑惑。理论与实践的搭配能更好的帮助大家学习，快去试试吧！若想继续学习更多相关知识，请继续关注编程网网站，小编会继续努力为大家带来更多实用的文章！
免责声明：
① 本站未注明“稿件来源”的信息均来自网络整理。其文字、图片和音视频稿件的所属权归原作者所有。本站收集整理出于非商业性的教育和科研之目的，并不意味着本站赞同其观点或证实其内容的真实性。仅作为临时的测试数据，供内部测试之用。本站并未授权任何人以任何方式主动获取本站任何信息。
② 本站未注明“稿件来源”的临时测试数据将在测试完成后最终做删除处理。有问题或投稿请发送至: 邮箱/279061341@qq.com QQ/279061341
