解释概述

人脸检测是指在图像或视频中检测出人脸的位置和大小。这个过程通常使用计算机视觉技术来实现。计算机会根据一些特征，比如肤色、面部轮廓等，来判断图像中是否存在人脸，并且确定人脸的位置和大小。人脸检测在很多应用中都有广泛的应用，比如人脸识别、视频监控等。

人脸识别是指在已知的人脸数据库中匹配并识别出一张或多张人脸。这个过程通常使用人工智能和机器学习技术来实现。计算机会通过学习已知的人脸数据库中的特征，比如面部轮廓、眼睛位置等，来识别新的人脸。人脸识别在安全领域、社交媒体等方面都有广泛的应用。

Robotpipe

robotpipe Python Library is an artificial intelligence learning library, including mainstream advanced artificial intelligence libraries, such as InsightFace, yolov8, paddleocr, etc.

本项目具备如下推理功能：

人脸检测
人脸识别
人脸年龄与性别识别
人脸106个特征点
换脸

B站演示地址

安装

安装opencv

pip install opencv-python-headless==4.6.0.66
pip install opencv-python==4.6.0.66
pip install opencv-contrib-python==4.6.0.66

安装onnxruntime

pip install onnxruntime

若你的电脑支持gpu,可以进入如下安装

pip install onnxruntime-gpu

pip安装

pip install robotpipe

若安装失败，请指定安装路径

pip install robotpipe -i https://pypi.tuna.tsinghua.edu.cn/simple

导入模型文件

在用户目录下新建.robotpipe文件夹

C:\Users\KAI\.robotpipe\models
在models里面放入权重文件

从百度网盘中下载权重文件,并解压释放进去

功能演示

人脸检测

人脸检测与识别 - 图2

from robotpipe import FaceDetection
from robotpipe import draw_faces
import cv2 as cv
import numpy as np
if __name__ == '__main__':
    face_detection = FaceDetection()
    img = cv.imread('images/trump1.jpg')
    faces = face_detection.predict(img)
    dst = draw_faces(img,faces)
    cv.imshow('img',dst)
    cv.waitKey(0)

人脸识别

人脸检测与识别 - 图4

from robotpipe import FaceRecognition
import cv2 as cv
import numpy as np
if __name__ == '__main__':
    face_recognition = FaceRecognition()
    trump1 = cv.imread('images/trump1.jpg')
    trump2 = cv.imread('images/trump2.jpg')
    # 获取人脸特征
    feats1 = face_recognition.predict(trump1)
    feats2 = face_recognition.predict(trump2)
    sim = face_recognition.compute_sim(feats1,feats2)
    # 根据相似度,输出结果
    if sim<0.2:
        conclu = 'they are not the same'
    elif sim>=0.2 and sim<0.28:
        conclu = 'they are looks like the same people'
    else:
        conclu = 'they are the same people'
    print(sim,conclu)
    cv.imshow('trump1',trump1)
    cv.imshow('trump2',trump2)
    cv.waitKey(0)

人脸年龄与性别识别

from robotpipe import FaceAttribute
from robotpipe import draw_faces
import cv2 as cv
import numpy as np
if __name__ == '__main__':
    face_attr = FaceAttribute()
    img = cv.imread('images/trump1.jpg')
    # 获取人脸特征
    faces = face_attr.predict(img)
    # 1 表示男性， 0，表示女性
    print(faces[0])
    dst = draw_faces(img,faces)
    cv.imshow("images/1.jpg",dst)
    cv.waitKey(0)

人脸106个特征点

from robotpipe import FaceLandmark
import cv2 as cv
import numpy as np
from heimarobot import draw_faces,draw_landmarks
if __name__ == '__main__':
    face_attr = FaceLandmark()
    trump1 = cv.imread('images/trump1.jpg')
    # 获取人脸特征106个特征点
    faces = face_attr.predict(trump1)
    retimg = draw_landmarks(trump1,faces)
    cv.imshow('trump1_landmark',retimg)
    cv.waitKey(0)

换脸

from robotpipe import FaceSwapper
import cv2 as cv
import numpy as np
if __name__ == '__main__':
    face_swapper = FaceSwapper()
    trump1 = cv.imread('images/trump1.jpg')
    target = cv.imread('images/1.jpg')
    sourceFace = face_swapper.get_source_face(trump1)
    dst = face_swapper.predict(sourceFace, target)
    cv.imshow("trump1",trump1)
    cv.imshow("1",target)
    cv.imshow("dst", dst)
    cv.waitKey()

若运行代码出现一下错误，则是opencv-python-headless与opencv-python版本不匹配的原因

cv2.error: OpenCV(4.7.0) D:\a\opencv-python\opencv-python\opencv\modules\highgui\src\window.cpp:1272: error: (-2:Unspecified error) The function is not implemented. Rebuild the library with Windows, GTK+ 2.x or Cocoa support. If you are on Ubuntu or Debian, install libgtk2.0-dev and pkg-config, then re-run cmake or configure script in function 'cvShowImage'

卸载opencv-python-headless,重新安装

pip install opencv-python-headless==4.6.0.66
pip install opencv-python==4.6.0.66
pip install opencv-contrib-python==4.6.0.66

本项目参考以下工程:

InsightFace
PaddleOCR
yolov8