百度api实现实时摄像头人脸识别

大家好，我是爱码士_秃头兔子。一名刚毕业从事计算机视觉的新人。如果文章对你有所帮助，希望能点个赞支持一下，谢谢~

本文是利用百度api接口和openCV库实现的摄像头实时人脸识别搜索，落地性不强，但应付毕业设计或者课程设计足够了，之前看同学花了老大一笔买了个这种程序，太亏了，我直接写篇文章吧，有缘人可以直接改改用。

创建百度智能云账号并登录

上一篇静态人脸识别已经写过详细教学了，不会的朋友可以看上一篇静态m：n搜索教程，前面东西都是一样的创建智能云账号，进入控制台，左侧列表展开选择人脸识别，领取免费资源，创建应用，创建人脸库。创建用户组，创建用户，添加用户人脸信息。
在此不再啰嗦，直接上硬货。

调用api实现

因为api接口要求输入的图片格式必须是Base64，所以我们要把视频抽帧出来并转化为base64编码格式。

def frame2base64(frame):
    img = Image.fromarray(frame) #将每一帧转为Image
    output_buffer = BytesIO()  # 创建一个BytesIO
    img.save(output_buffer, format='JPEG')  # 写入output_buffer
    byte_data = output_buffer.getvalue()  # 在内存中读取
    base64_data = base64.b64encode(byte_data)  # 转为BASE64
    return base64_data  # 转码成功 返回base64编码`` 

在人脸框上如果想标注中文是不能直接用cv2.putText（）的，这样如果标注的中文直接显示？？？
所以需要写一个函数：

def cv2ImgAddText(img, text, left, top, textColor=(0, 255, 0), textSize=20):
    if (isinstance(img, np.ndarray)):  # 判断是否OpenCV图片类型
        img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    # 创建一个可以在给定图像上绘图的对象
    draw = ImageDraw.Draw(img)
    # 字体的格式
    fontStyle = ImageFont.truetype("simsun.ttc", textSize, encoding="utf-8")
    # 绘制文本
    draw.text((left, top), text, textColor, font=fontStyle)
    # 转换回OpenCV格式
    return cv2.cvtColor(np.asarray(img), cv2.COLOR_RGB2BGR)

这样就可以显示中文了。

完整代码

import cv2
from aip import AipFace
from io import BytesIO
import base64
from PIL import Image,ImageDraw, ImageFont
import threading
import numpy as np

def cv2ImgAddText(img, text, left, top, textColor=(0, 255, 0), textSize=20):
    if (isinstance(img, np.ndarray)):  # 判断是否OpenCV图片类型
        img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    # 创建一个可以在给定图像上绘图的对象
    draw = ImageDraw.Draw(img)
    # 字体的格式
    fontStyle = ImageFont.truetype("simsun.ttc", textSize, encoding="utf-8")
    # 绘制文本
    draw.text((left, top), text, textColor, font=fontStyle)
    # 转换回OpenCV格式
    return cv2.cvtColor(np.asarray(img), cv2.COLOR_RGB2BGR)

def frame2base64(frame):
    img = Image.fromarray(frame) #将每一帧转为Image
    output_buffer = BytesIO()  # 创建一个BytesIO
    img.save(output_buffer, format='JPEG')  # 写入output_buffer
    byte_data = output_buffer.getvalue()  # 在内存中读取
    base64_data = base64.b64encode(byte_data)  # 转为BASE64
    return base64_data  # 转码成功 返回base64编码

def process(image,ls):
    """ 调用人脸检测 """
    """ 如果有可选参数 """

#换成自己的appid  
    APP_ID = ''
    API_KEY = ''
    SECRET_KEY = ''
    groupIdList = ''
    imageType = 'BASE64'
    base64=frame2base64(image)
    # base64 = str(base64)
    base64 = str(base64,'UTF-8')
    options = {}
    options["face_field"] = "age,gender,emotion"
    options["max_face_num"] = 10
    options["face_type"] = "LIVE"
    options["match_threshold"] = 0
    #options["liveness_control"] = "NORMAL"
    client = AipFace(APP_ID, API_KEY, SECRET_KEY)
    face_data=client.detect(base64, imageType, options)
    print(face_data)
    user_result = {}
    user_attribute = {}
    user_type = '访客'
    is_user = False
    if face_data['error_msg'] == 'SUCCESS':
        i = face_data['result']['face_list'][0]
        if i['face_probability']>0.8:
            is_user = True
            age=str(i['age'])
            gender = str('男' if i['gender']['type']== 'male' else '女')
            emotion = str(i['emotion']['type'])
            if emotion == 'disgust':
                emotion = 'angry'
            elif emotion == 'pouty':
                emotion = 'sad'
            elif emotion == 'grimace':
                emotion = 'happy'
            else:
                emotion = 'neutral'
            ##shibie
            # groupIdList = 'group_1'
            """ 如果有可选参数 """
            options = {}
            options["max_face_num"] = 10
            options["match_threshold"] = 0
            options["quality_control"] = "LOW"
            options["liveness_control"] = "LOW"
            options["max_user_num"] = 7
            # json1 = client.multiSearch(base64, imageType, groupIdList, options)
            json1 = client.multiSearch(base64, imageType, groupIdList, options)
            print(json1)
            face_num = face_data['result']['face_num']
            for i in range(face_num):
                x = max(int(face_data['result']['face_list'][i]['location']['left']), 0)
                y = max(int(face_data['result']['face_list'][i]['location']['top']), 0)
                width = int(face_data['result']['face_list'][i]['location']['width'])
                height = int(face_data['result']['face_list'][i]['location']['height'])
                cv2.rectangle(image, (x, y), (x + width, y + height), (0, 0, 255), 2)
                if json1['error_msg']=='SUCCESS':
                    user_attribute = []
                    user_type = '员工'
                    user_result = {'is_user':is_user,'user_type':user_type,'user_attribute':user_attribute}

                    if json1['result']['face_list'][0]['user_list'][i]['score']>70:
                        print(json1['result']['face_list'][0]['user_list'][i]['user_id'])
                        image = cv2ImgAddText(image,json1['result']['face_list'][0]['user_list'][i]['user_id'],
                                              max(x - 20, 0), max(y - 20, 0), (255, 255, 255), 20)
                    else:

                        cv2.putText(image, f"{str(face_data['result']['face_list'][i]['age'])} {face_data['result']['face_list'][i]['gender']['type']}", (x, y), cv2.FONT_HERSHEY_SIMPLEX, 1.2, (255, 255, 255), 2)
                    ls.append(image)

        else:
            user_result = {'is_user':is_user}
            print (user_result)
            # return user_result
    else:
        user_result = {'is_user':is_user}
        print (user_result)
        # return user_result
    # return user_result

def main():
    # video_capture = cv2.VideoCapture(0)
    video_capture = cv2.VideoCapture('1.avi')
    # video_capture.set(30, 30)
    # video_capture.set(6, cv2.VideoWriter.fourcc('M', 'J', 'P', 'G'))
    # video_capture.set(3, 640)  # Width of the frames in the video stream.
    # video_capture.set(4, 480)  # Height of the frames in the video stream.
    # video_capture.set(5, 30) # Frame rate.
    video_capture.set(cv2.CAP_PROP_FRAME_WIDTH, 640)  # 宽度
    video_capture.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)  # 高度
    video_capture.set(cv2.CAP_PROP_FPS, 30)  # 帧数
    video_capture.set(cv2.CAP_PROP_BRIGHTNESS, 1)  # 亮度 1
    video_capture.set(cv2.CAP_PROP_CONTRAST, 40)  # 对比度 40
    video_capture.set(cv2.CAP_PROP_SATURATION, 50)  # 饱和度 50
    video_capture.set(cv2.CAP_PROP_HUE, 50)  # 色调 50
    video_capture.set(cv2.CAP_PROP_EXPOSURE, 50)  # 曝光 50

    while True:
        ls = []
        ret, frame = video_capture.read()
        t = threading.Thread(target=process,args=(frame,ls))
        t.start()
        t.join()
        frame = ls[0] if ls else frame
        cv2.imshow('wx',frame)
        if cv2.waitKey(1) & 0xFF == ord('Q'):
            cv2.destroyAllWindows()
            video_capture.release()
            break
if __name__ == "__main__":
    main()

代码改一下appid那块就可以用啦，因为调用 的百度开源模型，并没有太大技术。适合新手小白，应付一下做个简单的毕业设计或者课程设计也挺好的。哪里不明白或者有问题可以留言私信，不定时会登录解答的。