赞
踩
环境准备
- 需要安装Visual Studio (C++平台),否则dlib模块无法安装成功。
- pip install dlib
- pip install numpy
- pip install opencv-python
- pip install scikit-image
import dlib from skimage import io # 使用Dilb的正面人脸检测器frontal_face_detector detector = dlib.get_frontal_face_detector() # Dlib 的人脸检测模型 predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat") # 图片所在路径 # img = io.imread("x3.jpg") img = io.imread("img.png") # 生成Dlib的图像窗口 win = dlib.image_window() win.set_image(img) # 使用detector检测器来检测图像中的人脸 faces = detector(img, 1) print("人脸数:", len(faces)) for i, d in enumerate(faces): print("第", i+1, "个人脸的矩形框坐标:", "left:", d.left(), "right:", d.right, "top:", d.top(), "bottom:", d.bottom) # 绘制人脸脸部矩形框 win.add_overlay(faces) # 保持图像 dlib.hit_enter_to_continue()
img.png:
运行效果:
测试如果一张图片中有多张人脸的效果:
import dlib from skimage import io # 使用Dilb的正面人脸检测器frontal_face_detector detector = dlib.get_frontal_face_detector() # Dlib 的人脸检测模型 predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat") # 图片所在路径 # img = io.imread("zsm1.jpg") img = io.imread("img.png") # 生成Dlib的图像窗口 win = dlib.image_window() win.set_image(img) # 使用detector检测器来检测图像中的人脸 faces = detector(img, 1) print("人脸数:", len(faces)) for i, d in enumerate(faces): print("第", i+1, "个人脸的矩形框坐标:", "left:", d.left(), "right:", d.right, "top:", d.top(), "bottom:", d.bottom) # 使用predictor来计算面部轮廓关键点位置 shape = predictor(img, faces[i]) # 绘制面部轮廓矩形框 win.add_overlay(shape) # 绘制人脸脸部矩形框 win.add_overlay(faces) # 保持图像 dlib.hit_enter_to_continue()
运行效果:
多张人脸运行效果:
import dlib # pip install dlib import glob import numpy # pip install numpy import os import sys import cv2 # pip install opencv-python from skimage import io # pip install scikit-image # 编写一个人脸识别程序 if len(sys.argv) != 2: # 命令行参数 print("请检查参数是否正确") exit() current_path = os.getcwd() # 获取当前路径 # 1. 人脸关键点检测器 # premod = "\\model\\shape_predictor_68_face_landmarks.dat" premod = "shape_predictor_68_face_landmarks.dat" # predictor_path = current_path + premod predictor_path = premod # 2. 人脸识别模型 # recmod = "\\model\\dlib_face_recognition_resnet_model_v1.dat" recmod = "dlib_face_recognition_resnet_model_v1.dat" # face_rec_model_path = current_path + recmod face_rec_model_path = recmod # 3. 备选人脸文件夹 faces_folder_path = "face_data1" # 4. 需识别的人脸 img_path = sys.argv[1] # img_path = "face_data1/img2.png" # 5. 加载正脸检测器 detector = dlib.get_frontal_face_detector() # 6. 加载人脸关键点检测器 sp = dlib.shape_predictor(predictor_path) # 7. 加载人脸识别模型 facerec = dlib.face_recognition_model_v1(face_rec_model_path) # 8. 加载显示人脸窗体 win = dlib.image_window() # 候选人脸描述子list descriptors = [] # 9. 对文件夹下的每一个人脸进行 # (1)人脸检测 # (2)关键点检测 # (3)描述子提取 # for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")): for f in glob.glob(os.path.join(faces_folder_path, "*.png")): print("Processing file: {}".format(f)) img = io.imread(f) win.clear_overlay() win.set_image(img) # (1)人脸检测 dets = detector(img, 1) print("Number of faces detected: {}".format(len(dets))) for k, d in enumerate(dets): # (2)关键点检测 shape = sp(img, d) # 画出人脸区域和关键点 win.clear_overlay() win.add_overlay(d) win.add_overlay(shape) # (3)描述子提取,128维向量 face_descriptor = facerec.compute_face_descriptor(img, shape) # 转换为numpy array v = numpy.array(face_descriptor) descriptors.append(v) # 10. 对需识别人脸进行同样处理 # 提取描述子,不再注释 img = io.imread(img_path) dets = detector(img, 1) adist = [] for k, d in enumerate(dets): shape = sp(img, d) face_descriptor = facerec.compute_face_descriptor(img, shape) d_test = numpy.array(face_descriptor) # 计算欧氏距离(什么是欧式距离:https://wiki.mbalib.com/wiki/%E6%AC%A7%E5%87%A0%E9%87%8C%E5%BE%97%E8%B7%9D%E7%A6%BB) # 即两点(正数)之间最短距离 for i in descriptors: dist_ = numpy.linalg.norm(i - d_test) adist.append(dist_) # 11. 候选人名单 candidate = ['jiejie', 'jobs', 'meimei', 'zsm1', 'zsm2', 'zsm3'] c_d = [] # 12. 候选人和距离组成一个dict c_d = dict(zip(candidate, adist)) print(c_d) cd_sorted = sorted(c_d.items(), key=lambda d:d[1]) print("\n 该照片上的人是:", cd_sorted[0][0]) # note: 模型下载地址:http://dlib.net/files/dlib_face_recognition_resnet_model_v1.dat.bz2 # 运行方法:python 11.6.py img.png # 运行方法:python 11.6.py img1.png
Copyright © 2003-2013 www.wpsshop.cn 版权所有,并保留所有权利。