用OpenCV和Dlib进行人脸颜值预测-解析

参考：
1、http://blog.csdn.net/wc781708249/article/details/78604740
2、http://blog.csdn.net/xingchenbingbuyu/article/details/52804013

---

数据下载：http://www.hcii-lab.net/data/SCUT-FBP/EN/introduce.html

---

图片与标签转成pickle文件

# -*- coding: UTF-8 -*-"""生成图像数据和标签数据下载：http://www.hcii-lab.net/data/SCUT-FBP/EN/introduce.html"""import csvimport pandas as pdimport numpy as npimport osimport cv2import globfrom sklearn import preprocessingfrom sklearn import decompositionfile_path="./Data_Collection/*.jpg"img_files=glob.glob(file_path)img_names=[]imgs=[]for img_file in img_files:    img_name = img_file.split('\\')[-1].split('.')[0].split('-')[-1]  # Linux 为'/'    img_names.append(img_name) # 记录图片名 后续对应标签    img=cv2.imread(img_file,0) # 简单点取灰度图    img=cv2.resize(img,(480,480)) # 尺寸统一    # # cv2.normalize(img,img,1.0,0.0,cv2.NORM_MINMAX) # 归一到0~1    img=img.flatten() # 按行展成一行 即每一行代表一张图片    imgs.append(img)imgs=np.array(imgs,np.float32)  # 转成 500x230400# 数据归一化normalizer = preprocessing.Normalizer(norm='l2').fit(imgs)imgs=normalizer.transform(imgs)# pca降维pca = decomposition.PCA(n_components=20,svd_solver='randomized',whiten=True).fit(imgs)imgs = pca.transform(imgs)# 提取图像对应的标签labels_path="./Rating_Collection/Attractiveness label.xlsx"labels=pd.read_excel(labels_path)labels=labels._values[:,:2]labels=dict(labels)labels2=[]# 数据与标签对应[labels2.append(labels[int(img_name)]) for img_name in img_names]# 现在 imgs 与 labels2 是对应的labels2=np.array(labels2,np.float32)[:,np.newaxis]data=np.hstack((imgs,labels2))# 保存到文件中pd.to_pickle(data,'./data.pkl')

---

加载pkl文件训练模型

# -*- coding: UTF-8 -*-import pandas as pdimport numpy as npimport argparseimport matplotlib.pyplot as pltfrom sklearn import decompositionfrom sklearn import linear_modelfrom sklearn.ensemble import RandomForestRegressorfrom sklearn import svmfrom sklearn import gaussian_processfrom sklearn.externals import joblib# 加载数据data=pd.read_pickle('./data.pkl')# 分成train_datas 与 test_datasnp.random.shuffle(data) # 随机打乱数据train_datas=data[0:-50]test_datas=data[-50:]# regr = linear_model.LinearRegression()# regr = svm.SVR()# regr = gaussian_process.GaussianProcess(theta0=1e-2, thetaL=1e-4, thetaU=1e-1)regr = RandomForestRegressor(n_estimators=50, max_depth=None, min_samples_split=1., random_state=0)regr = regr.fit(train_datas[:,:-1], train_datas[:,-1])#一定要把compress设为true或者其他的值，没有设置会输出很多的*.npyjoblib.dump(regr, './my_face_rating.pkl',compress=1)print("Generate Model Successfully!")

人脸关键点提取

参考：https://github.com/LiuXiaolong19920720/predict-facial-attractiveness/blob/master/source/getLandmarks.py

# -*- coding: UTF-8 -*-import cv2import dlibimport numpyimport sysPREDICTOR_PATH = "./data/shape_predictor_68_face_landmarks.dat"# 1.使用dlib自带的frontal_face_detector作为我们的人脸提取器detector = dlib.get_frontal_face_detector()# 2.使用官方提供的模型构建特征提取器predictor = dlib.shape_predictor(PREDICTOR_PATH)class NoFaces(Exception):    passim = cv2.imread("./image/girls.jpg")# 3.使用detector进行人脸检测 rects为返回的结果rects = detector(im, 1)# 4.输出人脸数，rects的元素个数即为脸的个数if len(rects) >= 1:    print("{} faces detected".format(len(rects)))if len(rects) == 0:    raise NoFacesf = open('./landmarks.txt', 'w')for i in range(len(rects)):    # 5.使用predictor进行人脸关键点识别    landmarks = numpy.matrix([[p.x, p.y] for p in predictor(im, rects[i]).parts()])    im = im.copy()    # 使用enumerate 函数遍历序列中的元素以及它们的下标    for idx, point in enumerate(landmarks):        pos = (point[0, 0], point[0, 1])        f.write(str(point[0, 0]))        f.write(',')        f.write(str(point[0, 1]))        f.write(',')        # cv2.putText(im,str(idx),pos,        # fontFace=cv2.FONT_HERSHEY_SCRIPT_SIMPLEX,        # fontScale=0.4,        # color=(0,0,255))        # 6.绘制特征点        cv2.circle(im, pos, 3, color=(0, 255, 0))    f.write('\n')print("landmarks,get!")cv2.namedWindow("im", 2)cv2.imshow("im", im)cv2.waitKey(1000)

特征生成

参考：https://github.com/LiuXiaolong19920720/predict-facial-attractiveness/blob/master/source/generateFeatures.py

import mathimport numpyimport itertoolsdef facialRatio(points):    x1 = points[0];    y1 = points[1];    x2 = points[2];    y2 = points[3];    x3 = points[4];    y3 = points[5];    x4 = points[6];    y4 = points[7];    dist1 = math.sqrt((x1-x2)**2 + (y1-y2)**2)    dist2 = math.sqrt((x3-x4)**2 + (y3-y4)**2)    ratio = dist1/dist2    return ratiodef generateFeatures(pointIndices1, pointIndices2, pointIndices3, pointIndices4, allLandmarkCoordinates):    size = allLandmarkCoordinates.shape    allFeatures = numpy.zeros((size[0], len(pointIndices1)))    for x in range(0, size[0]):        landmarkCoordinates = allLandmarkCoordinates[x, :]        ratios = [];        for i in range(0, len(pointIndices1)):            x1 = landmarkCoordinates[2*(pointIndices1[i]-1)]            y1 = landmarkCoordinates[2*pointIndices1[i] - 1]            x2 = landmarkCoordinates[2*(pointIndices2[i]-1)]            y2 = landmarkCoordinates[2*pointIndices2[i] - 1]            x3 = landmarkCoordinates[2*(pointIndices3[i]-1)]            y3 = landmarkCoordinates[2*pointIndices3[i] - 1]            x4 = landmarkCoordinates[2*(pointIndices4[i]-1)]            y4 = landmarkCoordinates[2*pointIndices4[i] - 1]            points = [x1, y1, x2, y2, x3, y3, x4, y4]            ratios.append(facialRatio(points))        allFeatures[x, :] = numpy.asarray(ratios)    return allFeaturesdef generateAllFeatures(allLandmarkCoordinates):    a = [18, 22, 23, 27, 37, 40, 43, 46, 28, 32, 34, 36, 5, 9, 13, 49, 55, 52, 58]    combinations = itertools.combinations(a, 4)    i = 0    pointIndices1 = [];    pointIndices2 = [];    pointIndices3 = [];    pointIndices4 = [];    for combination in combinations:        pointIndices1.append(combination[0])        pointIndices2.append(combination[1])        pointIndices3.append(combination[2])        pointIndices4.append(combination[3])        i = i+1        pointIndices1.append(combination[0])        pointIndices2.append(combination[2])        pointIndices3.append(combination[1])        pointIndices4.append(combination[3])        i = i+1        pointIndices1.append(combination[0])        pointIndices2.append(combination[3])        pointIndices3.append(combination[1])        pointIndices4.append(combination[2])        i = i+1    return generateFeatures(pointIndices1, pointIndices2, pointIndices3, pointIndices4, allLandmarkCoordinates)landmarks = numpy.loadtxt('./landmarks.txt', delimiter=',', usecols=range(136))featuresALL = generateAllFeatures(landmarks)numpy.savetxt('./my_features.txt', featuresALL, delimiter=',', fmt = '%.04f')print("Generate Feature Successfully!")#pointIndices1 = [20, 20, 45, 45]#pointIndices2 = [58, 9, 58, 58]#pointIndices3 = [5, 7, 5, 32]#pointIndices4 = [13, 13, 11, 36]#features = generateFeatures(pointIndices1, pointIndices2, pointIndices3, pointIndices4, landmarks)

颜值预测

参考：https://github.com/LiuXiaolong19920720/predict-facial-attractiveness/blob/master/source/myPredict.py

from sklearn.externals import joblibimport numpy as npfrom sklearn import decomposition#use your own pathclf = joblib.load('./my_face_rating.pkl')features = np.loadtxt(root + './features_ALL.txt', delimiter=',')my_features = np.loadtxt(root + './my_features.txt', delimiter=',')pca = decomposition.PCA(n_components=20)pca.fit(features)predictions = np.zeros([6,1])for i in range(0, 6):    features_test = features[i, :]    features_test = pca.transform(features_test)    #regr = linear_model.LinearRegression()    #regr.fit(features_train, ratings_train)    predictions[i] = clf.predict(features_test)#predictions = clf.predict(features)print (predictions)