使用到的庫： dlib+Opencv python版本： 3.8 編譯環(huán)境： Jupyter Notebook (Anaconda3) 提取特征點(diǎn)：請參考 首選抓取多張圖片，從中獲取特征數(shù)據(jù)集和平均特征值然后寫入 csv 文件 - 計(jì)算特征數(shù)據(jù)集的歐式距離作對比：首先使用Opencv庫將攝像頭中的人臉框出來，再將攝像頭中采取到的人臉特征值與數(shù)據(jù)集中的每個人的特征均值作對比，選取最接近（歐氏距離最小）的值，將其標(biāo)注為歐氏距離最小的數(shù)據(jù)集的人名 私信小編01即可獲取大量Python學(xué)習(xí)資源0.Dlib人臉特征檢測原理
請參考
在視頻流中抓取人臉特征，并保存為 256*256 大小的圖片文件共20張，這就是我們建立數(shù)據(jù)集的第一步，用來訓(xùn)練人臉識別。
不一定是256*256的尺寸，可以根據(jù)自己的需求來調(diào)整大小，圖片越大訓(xùn)練結(jié)果會愈加精確，但也會影響訓(xùn)練模型的時間。
其中：
光線：曝光和黑暗的圖片需手動剔除- 請使用同一個設(shè)備進(jìn)行數(shù)據(jù)采集，不同設(shè)備的攝像頭采集到的數(shù)據(jù)集會有出入- 這里采用的是從視頻流中進(jìn)行捕捉截圖，也可以自己準(zhǔn)備20張左右的人臉圖片
代碼：
import cv2 import dlib import os import sys import random # 存儲位置 output_dir = 'D:/No1WorkSpace/JupyterNotebook/Facetrainset/Num&Name' #這里填編號+人名 size = 256 #圖片邊長 if not os.path.exists(output_dir):
os.makedirs(output_dir) # 改變圖片的亮度與對比度 def relight(img, light=1, bias=0):
w = img.shape[1]
h = img.shape[0] #image = []
for i in range(0,w):
for j in range(0,h):
for c in range(3):
tmp = int(img[j,i,c]*light + bias)
if tmp > 255:
tmp = 255
elif tmp < 0:
tmp = 0
img[j,i,c] = tmp
return img #使用dlib自帶的frontal_face_detector作為我們的特征提取器 detector = dlib.get_frontal_face_detector() # 打開攝像頭 參數(shù)為輸入流，可以為攝像頭或視頻文件 camera = cv2.VideoCapture(0) #camera = cv2.VideoCapture('C:/Users/CUNGU/Videos/Captures/wang.mp4') index = 1 while True:
if (index <= 20):#存儲15張人臉特征圖像
print('Being processed picture %s' % index) # 從攝像頭讀取照片
success, img = camera.read() # 轉(zhuǎn)為灰度圖片
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 使用detector進(jìn)行人臉檢測
dets = detector(gray_img, 1)
for i, d in enumerate(dets):
x1 = d.top() if d.top() > 0 else 0
y1 = d.bottom() if d.bottom() > 0 else 0
x2 = d.left() if d.left() > 0 else 0
y2 = d.right() if d.right() > 0 else 0
face = img[x1:y1,x2:y2] # 調(diào)整圖片的對比度與亮度， 對比度與亮度值都取隨機(jī)數(shù)，這樣能增加樣本的多樣性
face = relight(face, random.uniform(0.5, 1.5), random.randint(-50, 50))
face = cv2.resize(face, (size,size))
cv2.imshow('image', face)
cv2.imwrite(output_dir+'/'+str(index)+'.jpg', face)
index += 1
key = cv2.waitKey(30) & 0xff
if key == 27:
break
else:
print('Finished!') # 釋放攝像頭 release camera
camera.release() # 刪除建立的窗口 delete all the windows
cv2.destroyAllWindows()
break運(yùn)行效果：
根據(jù)抓取的圖片和人臉識別模型->訓(xùn)練得到的20個的68個特征數(shù)據(jù)集以及1個平均特征值存入csv文件
每張圖片的68個特征數(shù)據(jù)集可以不用存取，他們只是中間量，計(jì)算平均值以后就可以拋棄了，這里把他們輸出出來只是為了方便學(xué)習(xí)。
代碼：
# 從人臉圖像文件中提取人臉特征存入 CSV # Features extraction from images and save into features_all.csv # return_128d_features() 獲取某張圖像的128D特征 # compute_the_mean() 計(jì)算128D特征均值 from cv2 import cv2 as cv2
import os
import dlib
from skimage import io
import csv
import numpy as np
# 要讀取人臉圖像文件的路徑 path_images_from_camera = "D:/No1WorkSpace/JupyterNotebook/Facetrainset/" # Dlib 正向人臉檢測器 detector = dlib.get_frontal_face_detector()
# Dlib 人臉預(yù)測器 predictor = dlib.shape_predictor("D:/No1WorkSpace/JupyterNotebook/model/shape_predictor_68_face_landmarks.dat")
# Dlib 人臉識別模型 # Face recognition model, the object maps human faces into 128D vectors face_rec = dlib.face_recognition_model_v1("D:/No1WorkSpace/JupyterNotebook/model/dlib_face_recognition_resnet_model_v1.dat")
# 返回單張圖像的 128D 特征 def return_128d_features(path_img):
img_rd = io.imread(path_img)
img_gray = cv2.cvtColor(img_rd, cv2.COLOR_BGR2RGB)
faces = detector(img_gray, 1)
print("%-40s %-20s" % ("檢測到人臉的圖像 / image with faces detected:", path_img), '\n')
# 因?yàn)橛锌赡芙叵聛淼娜四樤偃z測，檢測不出來人臉了
# 所以要確保是 檢測到人臉的人臉圖像 拿去算特征
if len(faces) != 0:
shape = predictor(img_gray, faces[0])
face_descriptor = face_rec.compute_face_descriptor(img_gray, shape)
else:
face_descriptor = 0
print("no face")
return face_descriptor
# 將文件夾中照片特征提取出來, 寫入 CSV def return_features_mean_personX(path_faces_personX):
features_list_personX = []
photos_list = os.listdir(path_faces_personX)
if photos_list:
for i in range(len(photos_list)):
with open("D:/No1WorkSpace/JupyterNotebook/feature/featuresGiao"+str(i)+".csv", "w", newline="") as csvfile:
writer = csv.writer(csvfile)
# 調(diào)用return_128d_features()得到128d特征
print("%-40s %-20s" % ("正在讀的人臉圖像 / image to read:", path_faces_personX + "/" + photos_list[i]))
features_128d = return_128d_features(path_faces_personX + "/" + photos_list[i])
print(features_128d)
writer.writerow(features_128d)
# 遇到?jīng)]有檢測出人臉的圖片跳過
if features_128d == 0:
i += 1
else:
features_list_personX.append(features_128d)
else:
print("文件夾內(nèi)圖像文件為空 / Warning: No images in " + path_faces_personX + '/', '\n')
# 計(jì)算 128D 特征的均值
# N x 128D -> 1 x 128D
if features_list_personX:
features_mean_personX = np.array(features_list_personX).mean(axis=0)
else:
features_mean_personX = '0'
return features_mean_personX
# 讀取某人所有的人臉圖像的數(shù)據(jù) people = os.listdir(path_images_from_camera)
people.sort()
with open("D:/No1WorkSpace/JupyterNotebook/feature/features_all.csv", "w", newline="") as csvfile:
writer = csv.writer(csvfile)
for person in people:
print("##### " + person + " #####")
# Get the mean/average features of face/personX, it will be a list with a length of 128D
features_mean_personX = return_features_mean_personX(path_images_from_camera + person)
writer.writerow(features_mean_personX)
print("特征均值 / The mean of features:", list(features_mean_personX))
print('\n')
print("所有錄入人臉數(shù)據(jù)存入 / Save all the features of faces registered into: D:/myworkspace/JupyterNotebook/People/feature/features_all2.csv")如果要輸出每一張圖片的特征數(shù)據(jù)集，這里要用到Python的文件批量生成。
代碼運(yùn)行效果
通過計(jì)算特征數(shù)據(jù)集的 歐氏距離 作對比來識別人臉，取歐氏距離最小的數(shù)據(jù)集進(jìn)行匹配。
歐氏距離也稱歐幾里得距離或歐幾里得度量，是一個通常采用的距離定義，它是在m維空間中兩個點(diǎn)之間的真實(shí)距離。在二維和三維空間中的歐氏距離的就是兩點(diǎn)之間的距離。使用這個距離，歐氏空間成為度量空間。相關(guān)聯(lián)的范數(shù)稱為歐幾里得范數(shù)。較早的文獻(xiàn)稱之為畢達(dá)哥拉斯度量。二維空間公式：
代碼：
# 攝像頭實(shí)時人臉識別 import os
import dlib # 人臉處理的庫 Dlib import csv # 存入表格 import time
import sys
import numpy as np # 數(shù)據(jù)處理的庫 numpy from cv2 import cv2 as cv2 # 圖像處理的庫 OpenCv import pandas as pd # 數(shù)據(jù)處理的庫 Pandas # 人臉識別模型，提取128D的特征矢量 # face recognition model, the object maps human faces into 128D vectors # Refer this tutorial: http://dlib.net/python/index.html#dlib.face_recognition_model_v1 facerec = dlib.face_recognition_model_v1("D:/No1WorkSpace/JupyterNotebook/model/dlib_face_recognition_resnet_model_v1.dat")
# 計(jì)算兩個128D向量間的歐式距離 # compute the e-distance between two 128D features def return_euclidean_distance(feature_1, feature_2):
feature_1 = np.array(feature_1)
feature_2 = np.array(feature_2)
dist = np.sqrt(np.sum(np.square(feature_1 - feature_2)))
return dist
# 處理存放所有人臉特征的 csv path_features_known_csv = "D:/No1WorkSpace/JupyterNotebook/feature/features_all.csv"
csv_rd = pd.read_csv(path_features_known_csv, header=None)
# 用來存放所有錄入人臉特征的數(shù)組 # the array to save the features of faces in the database features_known_arr = []
# 讀取已知人臉數(shù)據(jù) # print known faces for i in range(csv_rd.shape[0]):
features_someone_arr = []
for j in range(0, len(csv_rd.loc[i, :])):
features_someone_arr.append(csv_rd.loc[i, :][j])
features_known_arr.append(features_someone_arr)
print("Faces in Database：", len(features_known_arr))
# Dlib 檢測器和預(yù)測器 # The detector and predictor will be used detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor('D:/No1WorkSpace/JupyterNotebook/model/shape_predictor_68_face_landmarks.dat')
# 創(chuàng)建 cv2 攝像頭對象 # cv2.VideoCapture(0) to use the default camera of PC, # and you can use local video name by use cv2.VideoCapture(filename) cap = cv2.VideoCapture(0)
# cap.set(propId, value) # 設(shè)置視頻參數(shù)，propId 設(shè)置的視頻參數(shù)，value 設(shè)置的參數(shù)值 cap.set(3, 480)
# cap.isOpened() 返回 true/false 檢查初始化是否成功 # when the camera is open while cap.isOpened():
flag, img_rd = cap.read()
kk = cv2.waitKey(1)
# 取灰度
img_gray = cv2.cvtColor(img_rd, cv2.COLOR_RGB2GRAY)
# 人臉數(shù) faces
faces = detector(img_gray, 0)
# 待會要寫的字體 font to write later
font = cv2.FONT_HERSHEY_COMPLEX
# 存儲當(dāng)前攝像頭中捕獲到的所有人臉的坐標(biāo)/名字
# the list to save the positions and names of current faces captured
pos_namelist = []
name_namelist = []
# 按下 q 鍵退出
# press 'q' to exit
if kk == ord('q'):
break
else:
# 檢測到人臉 when face detected
if len(faces) != 0:
# 獲取當(dāng)前捕獲到的圖像的所有人臉的特征，存儲到 features_cap_arr
# get the features captured and save into features_cap_arr
features_cap_arr = []
for i in range(len(faces)):
shape = predictor(img_rd, faces[i])
features_cap_arr.append(facerec.compute_face_descriptor(img_rd, shape))
# 遍歷捕獲到的圖像中所有的人臉
# traversal all the faces in the database
for k in range(len(faces)):
print("##### camera person", k+1, "#####")
# 讓人名跟隨在矩形框的下方
# 確定人名的位置坐標(biāo)
# 先默認(rèn)所有人不認(rèn)識，是 unknown
# set the default names of faces with "unknown"
name_namelist.append("unknown")
# 每個捕獲人臉的名字坐標(biāo) the positions of faces captured
pos_namelist.append(tuple([faces[k].left(), int(faces[k].bottom() + (faces[k].bottom() - faces[k].top())/4)]))
# 對于某張人臉，遍歷所有存儲的人臉特征
# for every faces detected, compare the faces in the database
e_distance_list = []
for i in range(len(features_known_arr)):
# 如果 person_X 數(shù)據(jù)不為空
if str(features_known_arr[i][0]) != '0.0':
print("with person", str(i + 1), "the e distance: ", end='')
e_distance_tmp = return_euclidean_distance(features_cap_arr[k], features_known_arr[i])
print(e_distance_tmp)
e_distance_list.append(e_distance_tmp)
else:
# 空數(shù)據(jù) person_X
e_distance_list.append(999999999)
# 找出最接近的一個人臉數(shù)據(jù)是第幾個
# Find the one with minimum e distance
similar_person_num = e_distance_list.index(min(e_distance_list))
print("Minimum e distance with person", int(similar_person_num)+1)
# 計(jì)算人臉識別特征與數(shù)據(jù)集特征的歐氏距離
# 距離小于0.4則標(biāo)出為可識別人物
if min(e_distance_list) < 0.4:
# 這里可以修改攝像頭中標(biāo)出的人名
# Here you can modify the names shown on the camera
# 1、遍歷文件夾目錄
folder_name = 'D:/No1WorkSpace/JupyterNotebook/Facetrainset/'
# 最接近的人臉
sum=similar_person_num+1
key_id=1 # 從第一個人臉數(shù)據(jù)文件夾進(jìn)行對比
# 獲取文件夾中的文件名:1wang、2zhou、3...
file_names = os.listdir(folder_name)
for name in file_names:
# print(name+'->'+str(key_id))
if sum ==key_id:
#winsound.Beep(300,500)# 響鈴：300頻率，500持續(xù)時間
name_namelist[k] = name[1:]#人名刪去第一個數(shù)字（用于視頻輸出標(biāo)識）
key_id += 1
# 播放歡迎光臨音效
#playsound('D:/myworkspace/JupyterNotebook/People/music/welcome.wav')
# print("May be person "+str(int(similar_person_num)+1))
# -----------篩選出人臉并保存到visitor文件夾------------
for i, d in enumerate(faces):
x1 = d.top() if d.top() > 0 else 0
y1 = d.bottom() if d.bottom() > 0 else 0
x2 = d.left() if d.left() > 0 else 0
y2 = d.right() if d.right() > 0 else 0
face = img_rd[x1:y1,x2:y2]
size = 64
face = cv2.resize(face, (size,size))
# 要存儲visitor人臉圖像文件的路徑
path_visitors_save_dir = "D:/No1WorkSpace/JupyterNotebook/KnownFacetrainset/"
# 存儲格式：2019-06-24-14-33-40wang.jpg
now_time = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
save_name = str(now_time)+str(name_namelist[k])+'.jpg'
# print(save_name)
# 本次圖片保存的完整url
save_path = path_visitors_save_dir+'/'+ save_name
# 遍歷visitor文件夾所有文件名
visitor_names = os.listdir(path_visitors_save_dir)
visitor_name=''
for name in visitor_names:
# 名字切片到分鐘數(shù)：2019-06-26-11-33-00wangyu.jpg
visitor_name=(name[0:16]+'-00'+name[19:])
# print(visitor_name)
visitor_save=(save_name[0:16]+'-00'+save_name[19:])
# print(visitor_save)
# 一分鐘之內(nèi)重復(fù)的人名不保存
if visitor_save!=visitor_name:
cv2.imwrite(save_path, face)
print('新存儲：'+path_visitors_save_dir+'/'+str(now_time)+str(name_namelist[k])+'.jpg')
else:
print('重復(fù)，未保存！')
else:
# 播放無法識別音效
#playsound('D:/myworkspace/JupyterNotebook/People/music/sorry.wav')
print("Unknown person")
# -----保存圖片-------
# -----------篩選出人臉并保存到visitor文件夾------------
for i, d in enumerate(faces):
x1 = d.top() if d.top() > 0 else 0
y1 = d.bottom() if d.bottom() > 0 else 0
x2 = d.left() if d.left() > 0 else 0
y2 = d.right() if d.right() > 0 else 0
face = img_rd[x1:y1,x2:y2]
size = 64
face = cv2.resize(face, (size,size))
# 要存儲visitor-》unknown人臉圖像文件的路徑
path_visitors_save_dir = "D:/No1WorkSpace/JupyterNotebook/UnKnownFacetrainset/"
# 存儲格式：2019-06-24-14-33-40unknown.jpg
now_time = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
# print(save_name)
# 本次圖片保存的完整url
save_path = path_visitors_save_dir+'/'+ str(now_time)+'unknown.jpg'
cv2.imwrite(save_path, face)
print('新存儲：'+path_visitors_save_dir+'/'+str(now_time)+'unknown.jpg')
# 矩形框
# draw rectangle
for kk, d in enumerate(faces):
# 繪制矩形框
cv2.rectangle(img_rd, tuple([d.left(), d.top()]), tuple([d.right(), d.bottom()]), (0, 255, 255), 2)
print('\n')
# 在人臉框下面寫人臉名字
# write names under rectangle
for i in range(len(faces)):
cv2.putText(img_rd, name_namelist[i], pos_namelist[i], font, 0.8, (0, 255, 255), 1, cv2.LINE_AA)
print("Faces in camera now:", name_namelist, "\n")
#cv2.putText(img_rd, "Press 'q': Quit", (20, 450), font, 0.8, (84, 255, 159), 1, cv2.LINE_AA)
cv2.putText(img_rd, "Face Recognition", (20, 40), font, 1, (0, 0, 255), 1, cv2.LINE_AA)
cv2.putText(img_rd, "Visitors: " + str(len(faces)), (20, 100), font, 1, (0, 0, 255), 1, cv2.LINE_AA)
# 窗口顯示 show with opencv
cv2.imshow("camera", img_rd)
# 釋放攝像頭 release camera cap.release()
# 刪除建立的窗口 delete all the windows cv2.destroyAllWindows()若直接使用本代碼，文件目錄弄成中文會亂碼
運(yùn)行效果：
圖中兩人的特征數(shù)據(jù)集均已被收集并錄入，所以可以識別出來，如果沒有被錄入的人臉就會出現(xiàn)unknown。
沒有吳京叔叔的數(shù)據(jù)集，所以他是陌生人
