FacenetPytorch人脸识别方案--基于米尔全志T527开发板

apt-get update

sudo apt-get install -y python3-dev python3-pip libopenblas-dev libssl-dev libffi-dev git cmake

# 克隆 PyTorch 源代码git clone --recursive https://github.com/pytorch/pytorch# 进入 PyTorch 目录cd pytorch# 安装 PyTorch (需要根据你的需求选择 CUDA 版本，如果不需要 GPU 支持则不需要 --cuda 参数)pip3 install --no-cache-dir torch -f https://download.pytorch.org/whl/torch_stable.html# 测试 PyTorch 安装python3 -c "import torch; print(torch.__version__)"

pip3 install facenet_pytorch

############face_demo.py#############################import cv2import torchfrom facenet_pytorch import MTCNN, InceptionResnetV1# 获得人脸特征向量def load_known_faces(dstImgPath, mtcnn, resnet):aligned = []knownImg = cv2.imread(dstImgPath) # 读取图片face = mtcnn(knownImg) # 使用mtcnn检测人脸，返回人脸数组if face is not None:aligned.append(face[0])aligned = torch.stack(aligned).to(device)with torch.no_grad():known_faces_emb = resnet(aligned).detach().cpu()# 使用ResNet模型获取人脸对应的特征向量print("\n人脸对应的特征向量为：\n", known_faces_emb)return known_faces_emb, knownImg# 计算人脸特征向量间的欧氏距离，设置阈值，判断是否为同一张人脸def match_faces(faces_emb, known_faces_emb, threshold):isExistDst = Falsedistance = (known_faces_emb[0] - faces_emb[0]).norm().item()print("\n两张人脸的欧式距离为：%.2f" % distance)if (distance < threshold):isExistDst = Truereturn isExistDstif __name__ == '__main__':# help(MTCNN)# help(InceptionResnetV1)# 获取设备device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')# mtcnn模型加载设置网络参数，进行人脸检测mtcnn = MTCNN(min_face_size=12, thresholds=[0.2, 0.2, 0.3],keep_all=True, device=device)# InceptionResnetV1模型加载用于获取人脸特征向量resnet = InceptionResnetV1(pretrained='vggface2').eval().to(device)MatchThreshold = 0.8 # 人脸特征向量匹配阈值设置known_faces_emb, _ = load_known_faces('yz.jpg', mtcnn, resnet) # 已知人物图faces_emb, img = load_known_faces('yz1.jpg', mtcnn, resnet) # 待检测人物图isExistDst = match_faces(faces_emb, known_faces_emb, MatchThreshold) # 人脸匹配print("设置的人脸特征向量匹配阈值为：", MatchThreshold)if isExistDst:boxes, prob, landmarks = mtcnn.detect(img, landmarks=True)print('由于欧氏距离小于匹配阈值，故匹配')else:print('由于欧氏距离大于匹配阈值，故不匹配')

import matplotlib.pyplot as pltimport numpy as npimport torchimport torch.backends.cudnn as cudnnfrom nets.facenet import Facenet as facenetfrom utils.utils import preprocess_input, resize_image, show_config#--------------------------------------------## 使用自己训练好的模型预测需要修改2个参数# model_path和backbone需要修改！#--------------------------------------------#class Facenet(object):_defaults = {#--------------------------------------------------------------------------## 使用自己训练好的模型进行预测要修改model_path，指向logs文件夹下的权值文件# 训练好后logs文件夹下存在多个权值文件，选择验证集损失较低的即可。# 验证集损失较低不代表准确度较高，仅代表该权值在验证集上泛化性能较好。#--------------------------------------------------------------------------#"model_path" : "model_data/facenet_mobilenet.pth",#--------------------------------------------------------------------------## 输入图片的大小。#--------------------------------------------------------------------------#"input_shape" : [160, 160, 3],#--------------------------------------------------------------------------## 所使用到的主干特征提取网络#--------------------------------------------------------------------------#"backbone" : "mobilenet",#-------------------------------------------## 是否进行不失真的resize#-------------------------------------------#"letterbox_image" : True,#-------------------------------------------## 是否使用Cuda# 没有GPU可以设置成False#-------------------------------------------#"cuda" : False,}@classmethoddef get_defaults(cls, n):if n in cls._defaults:return cls._defaults[n]else:return "Unrecognized attribute name '" + n + "'"#---------------------------------------------------## 初始化Facenet#---------------------------------------------------#def __init__(self, **kwargs):self.__dict__.update(self._defaults)for name, value in kwargs.items():setattr(self, name, value)self.generate()show_config(**self._defaults)def generate(self):#---------------------------------------------------## 载入模型与权值#---------------------------------------------------#print('Loading weights into state dict...')device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')self.net = facenet(backbone=self.backbone, mode="predict").eval()self.net.load_state_dict(torch.load(self.model_path, map_location=device), strict=False)print('{} model loaded.'.format(self.model_path))if self.cuda:self.net = torch.nn.DataParallel(self.net)cudnn.benchmark = Trueself.net = self.net.cuda()#---------------------------------------------------## 检测图片#---------------------------------------------------#def detect_image(self, image_1, image_2):#---------------------------------------------------## 图片预处理，归一化#---------------------------------------------------#with torch.no_grad():image_1 = resize_image(image_1, [self.input_shape[1], self.input_shape[0]], letterbox_image=self.letterbox_image)image_2 = resize_image(image_2, [self.input_shape[1], self.input_shape[0]], letterbox_image=self.letterbox_image)photo_1 = torch.from_numpy(np.expand_dims(np.transpose(preprocess_input(np.array(image_1, np.float32)), (2, 0, 1)), 0))photo_2 = torch.from_numpy(np.expand_dims(np.transpose(preprocess_input(np.array(image_2, np.float32)), (2, 0, 1)), 0))if self.cuda:photo_1 = photo_1.cuda()photo_2 = photo_2.cuda()#---------------------------------------------------## 图片传入网络进行预测#---------------------------------------------------#output1 = self.net(photo_1).cpu().numpy()output2 = self.net(photo_2).cpu().numpy()#---------------------------------------------------## 计算二者之间的距离#---------------------------------------------------#l1 = np.linalg.norm(output1 - output2, axis=1)plt.subplot(1, 2, 1)plt.imshow(np.array(image_1))plt.subplot(1, 2, 2)plt.imshow(np.array(image_2))plt.text(-12, -12, 'Distance:%.3f' % l1, ha='center', va= 'bottom',fontsize=11)plt.show()return l1

from PIL import Imagefrom facenet import Facenetif __name__ == "__main__":model = Facenet()while True:image_1 = input('Input image_1 filename:')try:image_1 = Image.open(image_1)except:print('Image_1 Open Error! Try again!')continueimage_2 = input('Input image_2 filename:')try:image_2 = Image.open(image_2)except:print('Image_2 Open Error! Try again!')continueprobability = model.detect_image(image_1,image_2)print(probability)