1 創(chuàng)建環(huán)境

conda create -n face python=3.7
conda activate face
pip install opencv-python
pip install pillow
pip install opencv-contrib-python

2 準(zhǔn)備工作

2.1新建members.csv文件

文件內(nèi)容依次是id,First_name,Last_name,如圖:

2.2新建face文件夾

里面存放采集的人臉信息，用于訓(xùn)練

k = cv2.waitKey(1)
        if k == 27:  # 通過esc鍵退出攝像
            break
        elif count >= 200:  # 得到n個(gè)樣本后退出攝像（樣本越大，精度越高，但采集信息的時(shí)間也越長(zhǎng)）
            break

2.3注意事項(xiàng)

1. 臉部識(shí)別特征模塊

Path路徑為你創(chuàng)建環(huán)境下的cv2包中haarcascade_frontalface_default.xml對(duì)應(yīng)的地址
只加了opencv中臉部特征，沒加眼部識(shí)別。（cv2包中還有眼部特征）

Path = r"C:\Users\11931\.conda\envs\face1\Lib\site-packages\cv2\data\haarcascade_frontalface_default.xml"

2. 調(diào)用攝像頭
   手機(jī)下載一個(gè)ip攝像頭，將電腦和手機(jī)連一個(gè)無線網(wǎng)（建議電腦連手機(jī)熱點(diǎn)），將參數(shù)改為手機(jī)ip地址。
   

3 源碼

import cv2
import os
import numpy as np
from PIL import Image
import datetime
import csv


# 調(diào)用筆記本內(nèi)置攝像頭，所以參數(shù)為0，如果有其他的攝像頭可以調(diào)整參數(shù)為1，2
Path = r"C:\Users\11931\.conda\envs\face1\Lib\site-packages\cv2\data\haarcascade_frontalface_default.xml"
face_detector = cv2.CascadeClassifier(Path)
names = []
zh_name = []
with open("members.csv", "r", encoding='UTF-8') as csv_file:
    reader = csv.reader(csv_file)
    for item in reader:
        # print(item)
        names.append(item[2])
        zh_name.append(item[1])
    # print (zh_name)


def data_collection():
    cap = cv2.VideoCapture(0, cv2.CAP_DSHOW)
    # cv2.CAP_DSHOW是作為open調(diào)用的一部分傳遞標(biāo)志，還有許多其它的參數(shù)，而這個(gè)CAP_DSHOW是微軟特有的。
    face_id = input('\n 請(qǐng)輸入你的ID:')

    print('\n 數(shù)據(jù)初始化中，請(qǐng)直視攝像機(jī)錄入數(shù)據(jù)....')

    count = 0

    while True:

        # 從攝像頭讀取圖片

        sucess, img = cap.read()

        # 轉(zhuǎn)為灰度圖片

        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

        # 檢測(cè)人臉

        faces = face_detector.detectMultiScale(gray, 1.3, 5)

        for (x, y, w, h) in faces:
            cv2.rectangle(img, (x, y), (x + w, y + w), (255, 0, 0))
            count += 1
            # 保存圖像
            cv2.imwrite("11/Member." + str(face_id) + '.' + str(count) + '.jpg', gray[y: y + h, x: x + w])
            cv2.imshow('data collection', img)

        # 保持畫面的持續(xù)。

        k = cv2.waitKey(1)
        if k == 27:  # 通過esc鍵退出攝像
            break
        elif count >= 200:  # 得到n個(gè)樣本后退出攝像
            break
    cap.release()
    cv2.destroyAllWindows()


def face_training():
    # 人臉數(shù)據(jù)路徑
    path = './face'

    recognizer = cv2.face.LBPHFaceRecognizer_create()

    def getImagesAndLabels(path):
        imagePaths = [os.path.join(path, f) for f in os.listdir(path)]  # join函數(shù)將多個(gè)路徑組合后返回
        faceSamples = []
        ids = []
        for imagePath in imagePaths:
            PIL_img = Image.open(imagePath).convert('L')  # convert it to grayscale
            img_numpy = np.array(PIL_img, 'uint8')
            id = int(os.path.split(imagePath)[-1].split(".")[1])
            faces = face_detector.detectMultiScale(img_numpy)
            for (x, y, w, h) in faces:
                faceSamples.append(img_numpy[y:y + h, x: x + w])
                ids.append(id)
        return faceSamples, ids

    print('數(shù)據(jù)訓(xùn)練中')
    faces, ids = getImagesAndLabels(path)
    recognizer.train(faces, np.array(ids))

    recognizer.write(r'.\trainer.yml')
    # print("{0} faces trained. Exiting Program".format(len(np.unique(ids))))


def face_ientification():
    cap = cv2.VideoCapture(0)
    recognizer = cv2.face.LBPHFaceRecognizer_create()
    recognizer.read('./trainer.yml')
    faceCascade = cv2.CascadeClassifier(Path)
    font = cv2.FONT_HERSHEY_SIMPLEX

    idnum = 0
    global namess
    cam = cv2.VideoCapture(0)
    # 設(shè)置大小
    minW = 0.1 * cam.get(3)
    minH = 0.1 * cam.get(4)

    while True:
        ret, img = cam.read()
        # 圖像灰度處理
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

        # 將人臉用vector保存各個(gè)人臉的坐標(biāo)、大?。ㄓ镁匦伪硎荆?        faces = faceCascade.detectMultiScale(
            gray,
            scaleFactor=1.2,  # 表示在前后兩次相繼的掃描中，搜索窗口的比例系數(shù)
            minNeighbors=5,  # 表示構(gòu)成檢測(cè)目標(biāo)的相鄰矩形的最小個(gè)數(shù)(默認(rèn)為3個(gè))
            minSize=(int(minW), int(minH))  # minSize和maxSize用來限制得到的目標(biāo)區(qū)域的范圍
        )

        for (x, y, w, h) in faces:
            cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
            idnum, confidence = recognizer.predict(gray[y:y + h, x:x + w])

            if confidence < 100:
                namess = names[idnum]
                confidence = "{0}%".format(round(100 - confidence))
            else:
                namess = "unknown"
                confidence = "{0}%".format(round(100 - confidence))

            cv2.putText(img, str(namess), (x + 5, y - 5), font, 1, (0, 0, 255), 1)
            cv2.putText(img, str(confidence), (x + 5, y + h - 5), font, 1, (0, 0, 0), 1)  # 輸出置信度

        cv2.imshow(u'Identification punch', img)
        k = cv2.waitKey(10)
        if k == 13:
            theTime = datetime.datetime.now()
            # print(zh_name[idnum])
            strings = [str(zh_name[idnum]), str(theTime)]
            print(strings)
            with open("log.csv", "a", newline="") as csvFile:
                writer = csv.writer(csvFile)
                writer.writerow([str(zh_name[idnum]), str(theTime)])
        elif k == 27:
            cap.release()
            cv2.destroyAllWindows()
            break


while True:
    a = int(input("輸入1，錄入臉部，輸入2進(jìn)行識(shí)別打卡:"))
    if a == 1:
        data_collection()
        face_training()
    elif a == 2:
        face_ientification()

4 操作步驟

1. 錄入臉部 （輸入id,按照之前建立的csv文件序號(hào)操作）

2. 識(shí)別打卡（按下enter會(huì)錄入信息，esc退出）

文章來源地址http://www.zghlxwxcb.cn/news/detail-788555.html

3. 查看log（按下enter會(huì)錄入信息，esc退出）
   

到了這里，關(guān)于基于python+opencv的人臉識(shí)別打卡（手把手教你）的文章就介紹完了。如果您還想了解更多內(nèi)容，請(qǐng)?jiān)谟疑辖撬阉鱐OY模板網(wǎng)以前的文章或繼續(xù)瀏覽下面的相關(guān)文章，希望大家以后多多支持TOY模板網(wǎng)！