人脸识别具体案例(李智恩) 人脸识别具体案例(李智恩)

项目环境：python3.6
一、项目结构
文章图片

文章图片

二、数据集准备数据集准备分为两步：

获取图片.
提取人脸.

1、获取图片
首先可以利用爬虫，从百度图片上批量下载图片，但注意下载数据集所用的关键词不要和之后识别任务的关键词太接近，否则若有图片重合，就会产生“识别得很准”的错觉。下面的程序为爬虫部分，在name.txt文件中写好要搜索的关键词，即可使用。

# 爬虫部分，存放到 name + ‘文件'#############################################################################################if GET_PIC == 1:headers = {'Accept-Language': 'zh-CN,zh; q=0.8,zh-TW; q=0.7,zh-HK; q=0.5,en-US; q=0.3,en; q=0.2','Connection': 'keep-alive','User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Firefox/60.0','Upgrade-Insecure-Requests': '1'}A = requests.Session()A.headers = headerstm = int(input('请输入每类图片的下载数量 '))numPicture = tmline_list = []with open('./name.txt', encoding='utf-8') as file:line_list = [k.strip() for k in file.readlines()]# 用 strip()移除末尾的空格for word in line_list:url = 'https://image.baidu.com/search/flip?tn=baiduimage&ie=utf-8&word=' + word + '&pn='tot = Find(url, A)Recommend = recommend(url)# 记录相关推荐print('经过检测%s类图片共有%d张' % (word, tot))file = word + '文件'y = os.path.exists(file)if y == 1:print('该文件已存在，无需创建')else:os.mkdir(file)t = 0tmp = urlwhile t < numPicture:try:url = tmp + str(t)# result = requests.get(url, timeout=10)# 这里搞了下result = A.get(url, timeout=10, allow_redirects=False)print(url)except error.HTTPError as e:print('网络错误，请调整网络后重试')t = t + 60else:dowmloadPicture(result.text, word)t = t + 60numPicture = numPicture + tmprint('当前搜索结束，开始提取人脸')#############################################################################################

下载图片时要注意区分，将IU的图片放在一个文件夹下，Other的放在另一文件夹下。训练集和测试集都要如此。如下图所示：

文章图片

【人脸识别具体案例(李智恩)】每个文件夹内都是下图形式：

文章图片

在IU文件夹内图片如下所示：

文章图片

对于文件夹内文件的命名，可以利用以下这段程序，按顺序重命名。

import osraw_train_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/train/IU/'raw_train_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/train/Other/'raw_test_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/test/IU/'raw_test_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/test/Other/'raw_roots = [raw_train_root_1, raw_train_root_2, raw_test_root_1, raw_test_root_2]for path in raw_roots:# 获取该目录下所有文件，存入列表中fileList = os.listdir(path)n = 0for i in fileList:# 设置旧文件名（就是路径+文件名）oldname = path + os.sep + fileList[n]# os.sep添加系统分隔符# 设置新文件名newname = path + os.sep + str(n) + '.JPG'os.rename(oldname, newname)# 用os模块中的rename方法对文件改名print(oldname, '======>', newname)n += 1

2.提取人脸
提取人脸，需要用到一个人脸识别库face_recognition库。face_recognition库的下载步骤参考：
https://www.jb51.net/article/209870.htm
主要有三步，可以直接在anaconda的命令行界面复制使用：

pip install CMake -i https://pypi.douban.com/simple
pip install dlib==19.7.0 -i https://pypi.douban.com/simple
pip install face_recognition -i https://pypi.douban.com/simple

笔者已尝试，确实可用。
使用下述的函数就可以获得一张图片对应的人脸，返回值就是人脸图片。

# 找到图片中的人脸#############################################################################################def find_face(path):# Load the jpg file into a numpy arrayimage = face_recognition.load_image_file(path)# Find all the faces in the image using the default HOG-based model.# This method is fairly accurate, but not as accurate as the CNN model and not GPU accelerated.# See also: find_faces_in_picture_cnn.pyface_locations = face_recognition.face_locations(image) # 可以选择 model="cnn"if len(face_locations) == 0:return Noneelse:for face_location in face_locations:# Print the location of each face in this imagetop, right, bottom, left = face_location# You can access the actual face itself like this:face_image = image[top:bottom, left:right]pil_image = Image.fromarray(face_image)return pil_image#############################################################################################

对数据集进行操作之后，就可以获得处理后的人脸图片。之所以不用人物图训练，而是提取出人脸后再进行训练，是考虑到人物图像中干扰因素太多，且经过试验后发现识别的效果非常差，于是加入这个提取人脸的环节。对数据集的操作代码如下：

# 将训练集和测试集中的raw图片处理，提取出人脸图片#############################################################################################if __name__ == '__main__':# 主函数入口raw_train_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/train/IU/'raw_train_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/train/Other/'raw_test_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/test/IU/'raw_test_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/test/Other/'raw_roots = [raw_train_root_1, raw_train_root_2, raw_test_root_1, raw_test_root_2]img_raw_train_1 = os.listdir(raw_train_root_1)img_raw_train_2 = os.listdir(raw_train_root_2)img_raw_test_1 = os.listdir(raw_test_root_1)img_raw_test_2 = os.listdir(raw_test_root_2)img_raws = [img_raw_train_1, img_raw_train_2, img_raw_test_1, img_raw_test_2]new_path_train_1 = 'E:/Table/学习数据集/find_iu/data/processed/train/IU/'new_path_train_2 = 'E:/Table/学习数据集/find_iu/data/processed/train/Other/'new_path_test_1 = 'E:/Table/学习数据集/find_iu/data/processed/test/IU/'new_path_test_2 = 'E:/Table/学习数据集/find_iu/data/processed/test/Other/'new_paths = [new_path_train_1, new_path_train_2, new_path_test_1, new_path_test_2]for raw_root, img_raw, new_path in zip(raw_roots, img_raws, new_paths):n = 0for i in range(len(img_raw)):try:img = Image.open(raw_root + img_raw[i])except:print('a file error, continue')continueelse:img_train = find_face(raw_root + img_raw[i])if img_train == None:continueelse:# img_train.save(new_path + '%d.JPG'%n)# print(raw_root + img_raw[i])n += 1print('在%d张图片中，共找到%d张脸' % (len(img_raw), n))#############################################################################################

处理前的图片数据均存放在raw文件夹中，处理后的存放在processed文件夹中，如下图：

文章图片

两个文件夹的内部结构完全一样：

文章图片

文章图片

三、网络模型 1、图像处理
将图片裁剪为112×92大小，使用RGB图像，（这里试过用灰度图像，但好像效果不会更好，就放弃了），在对图片进行归一化处理。

data_transform = transforms.Compose([# transforms.Grayscale(num_output_channels=1),# 彩色图像转灰度图像num_output_channels默认1transforms.Resize(112),transforms.CenterCrop((112, 92)),# 中心裁剪为112*92transforms.ToTensor(),transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])# transforms.Normalize(mean=0.5, std=0.5)])

使用孪生神经网络(Siamese Network)

class SiameNetwork(nn.Module):def __init__(self):super(SiameNetwork, self).__init__()# input: h=112, w=92self.conv1 = torch.nn.Sequential(torch.nn.Conv2d(in_channels=3,# 输入单通道out_channels=16,# 16个3*3卷积核kernel_size=3,# 卷积核尺寸stride=2,# 卷积核滑动步长, 1的话图片大小不变，2的话会大小会变为(h/2)*(w/2)padding=1),# 边缘填充大小，如果要保持原大小，kernel_size//2torch.nn.BatchNorm2d(16),# 标准化，前面卷积后有16个图层torch.nn.ReLU()# 激活函数)# output: h=56, w=46self.conv2 = torch.nn.Sequential(torch.nn.Conv2d(16, 32, 3, 2, 1),torch.nn.BatchNorm2d(32),torch.nn.ReLU())# output: h=28, w=23self.conv3 = torch.nn.Sequential(torch.nn.Conv2d(32, 64, 3, 2, 1),torch.nn.BatchNorm2d(64),torch.nn.ReLU())# output: h=14, w=12self.conv4 = torch.nn.Sequential(torch.nn.Conv2d(64, 64, 2, 2, 0),torch.nn.BatchNorm2d(64),torch.nn.ReLU())# output: h=7, w=6self.mlp1 = torch.nn.Linear(7 * 6 * 64, 100)# 需要计算conv4的输出尺寸，每次卷积的输出尺寸(size - kernal + 2*padding)/stride + 1self.mlp2 = torch.nn.Linear(100, 10)def forward(self, x):x = self.conv1(x)x = self.conv2(x)x = self.conv3(x)x = self.conv4(x)x = self.mlp1(x.view(x.size(0), -1))# view展平x = self.mlp2(x)return x

四、具体代码 1.get_face.py

from PIL import Imageimport face_recognitionimport os# 找到图片中的人脸#############################################################################################def find_face(path):# Load the jpg file into a numpy arrayimage = face_recognition.load_image_file(path)# Find all the faces in the image using the default HOG-based model.# This method is fairly accurate, but not as accurate as the CNN model and not GPU accelerated.# See also: find_faces_in_picture_cnn.pyface_locations = face_recognition.face_locations(image) # 可以选择 model="cnn"if len(face_locations) == 0:return Noneelse:for face_location in face_locations:# Print the location of each face in this imagetop, right, bottom, left = face_location# You can access the actual face itself like this:face_image = image[top:bottom, left:right]pil_image = Image.fromarray(face_image)return pil_image############################################################################################## 将训练集和测试集中的raw图片处理，提取出人脸图片#############################################################################################if __name__ == '__main__':# 主函数入口raw_train_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/train/IU/'raw_train_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/train/Other/'raw_test_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/test/IU/'raw_test_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/test/Other/'raw_roots = [raw_train_root_1, raw_train_root_2, raw_test_root_1, raw_test_root_2]img_raw_train_1 = os.listdir(raw_train_root_1)img_raw_train_2 = os.listdir(raw_train_root_2)img_raw_test_1 = os.listdir(raw_test_root_1)img_raw_test_2 = os.listdir(raw_test_root_2)img_raws = [img_raw_train_1, img_raw_train_2, img_raw_test_1, img_raw_test_2]new_path_train_1 = 'E:/Table/学习数据集/find_iu/data/processed/train/IU/'new_path_train_2 = 'E:/Table/学习数据集/find_iu/data/processed/train/Other/'new_path_test_1 = 'E:/Table/学习数据集/find_iu/data/processed/test/IU/'new_path_test_2 = 'E:/Table/学习数据集/find_iu/data/processed/test/Other/'new_paths = [new_path_train_1, new_path_train_2, new_path_test_1, new_path_test_2]for raw_root, img_raw, new_path in zip(raw_roots, img_raws, new_paths):n = 0for i in range(len(img_raw)):try:img = Image.open(raw_root + img_raw[i])except:print('a file error, continue')continueelse:img_train = find_face(raw_root + img_raw[i])if img_train == None:continueelse:# img_train.save(new_path + '%d.JPG'%n)# print(raw_root + img_raw[i])n += 1print('在%d张图片中，共找到%d张脸' % (len(img_raw), n))#############################################################################################

2.find_iu.py

import torchimport torchvisionimport torch.nn as nnfrom torch.autograd import Variablefrom torchvision import datasets, transformsfrom torch.utils.data import DataLoaderimport cv2#opencv库，用于图片可视化import numpy as npimport osfrom utils import draw_resultfrom network import SiameNetworkfrom get_face import find_faceif __name__ == '__main__':# 主函数入口# 设置参数#############################################################################################path = 'E:/Table/学习数据集/find_iu/result/'# 存放和生成结果的路径标志epochs = 20#训练周期BATCH_SIZE = 16#批量样本大小NUM_WORKERS = 0############################################################################################## 数据处理#############################################################################################data_transform = transforms.Compose([# transforms.Grayscale(num_output_channels=1),# 彩色图像转灰度图像num_output_channels默认1transforms.Resize(112),transforms.CenterCrop((112, 92)),# 中心裁剪为112*92transforms.ToTensor(),transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])# transforms.Normalize(mean=0.5, std=0.5)])train_dataset = datasets.ImageFolder(root = r'E:/Table/学习数据集/find_iu/data/processed/train',transform = data_transform)test_dataset = datasets.ImageFolder(root = r'E:/Table/学习数据集/find_iu/data/processed/test',transform = data_transform)train_loader = DataLoader(train_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=NUM_WORKERS)test_loader = DataLoader(test_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=NUM_WORKERS)image, labels = next(iter(train_loader))#数据可视化img = torchvision.utils.make_grid(image, nrow = 10)img = img.numpy().transpose(1, 2, 0)cv2.imshow('img', img)#展示图像cv2.waitKey(0)#按下任一按键后开始工作print("data ready!")##############################################################################################配置设备、损失函数和优化器#############################################################################################device = torch.device('cuda')model = SiameNetwork().to(device)cost = torch.nn.CrossEntropyLoss()#定义损失函数，使用交叉熵optimizer = torch.optim.Adam(model.parameters(), lr=0.0008, weight_decay=0.001)#Adam优化器print("device ready!")##############################################################################################训练过程，训练周期由epochs决定#############################################################################################draw_epoch = []#记录训练阶段draw_loss = []#记录训练损失，用于绘制draw_train_acc = []#记录训练准确度，用于绘制draw_val_loss = []#记录测试损失，用于绘制draw_val_acc = []# 记录测试准确度，用于绘制for epoch in range(epochs):#训练过程sum_loss = 0.0sum_val_loss = 0.0train_correct = 0test_correct = 0for data in train_loader:inputs,labels = datainputs,labels = Variable(inputs).cuda(),Variable(labels).cuda()optimizer.zero_grad()#将上一batch梯度清零outputs = model(inputs)loss = cost(outputs, labels)loss.backward()#反向传播optimizer.step()_, id = torch.max(outputs.data, 1)sum_loss += loss.datatrain_correct += torch.sum(id == labels.data)for data in test_loader:# 模型测试inputs,labels = datainputs,labels = Variable(inputs).cuda(),Variable(labels).cuda()outputs = model(inputs)val_loss = cost(outputs, labels)_,id = torch.max(outputs.data, 1)sum_val_loss += val_loss.datatest_correct += torch.sum(id == labels.data)print('[%d,%d] train loss:%.03ftrain acc:%.03f%%'%(epoch + 1, epochs, sum_loss / len(train_loader), (100 * train_correct / len(train_dataset))))print('val loss:%.03fval acc:%.03f%%'%(sum_val_loss / len(test_loader), (100 * test_correct / len(test_dataset))))draw_epoch.append(epoch+1)# 用于后续画图的数据draw_loss.append(sum_loss / len(train_loader))draw_train_acc.append(100 * train_correct / len(train_dataset))draw_val_loss.append(sum_val_loss / len(test_loader))draw_val_acc.append(100 * test_correct / len(test_dataset))np.savetxt('%s/train_loss.txt'%(path), draw_loss, fmt="%.3f")# 保存损失数据np.savetxt('%s/train_acc.txt'%(path), draw_train_acc, fmt="%.3f")# 保存准确率数据np.savetxt('%s/val_loss.txt'%(path), draw_val_loss, fmt="%.3f")# 保存损失数据np.savetxt('%s/val_acc.txt'%(path), draw_val_acc, fmt="%.3f")# 保存准确率数据print("train ready!")##############################################################################################数据可视化#############################################################################################draw_result(draw_epoch, path)# 绘图函数print("draw ready!")##############################################################################################模型的存储和载入#############################################################################################torch.save(model.state_dict(), "parameter.pkl") #saveprint("save ready!")#############################################################################################

3.spider_iu.py

import reimport requestsfrom urllib import errorfrom bs4 import BeautifulSoupimport osimport torchfrom torch.autograd import Variablefrom torchvision import datasets, transformsfrom torch.utils.data import DataLoaderfrom network import SiameNetworkfrom utils import cv_imreadimport cv2from PIL import Imageimport shutilfrom get_face import find_face# 设置参数#############################################################################################GET_PIC = 0# 1 执行这步，0 不执行GET_FACE = 0GET_IU = 1#############################################################################################num = 0numPicture = 0file = ''List = []# 爬虫所用函数#############################################################################################def Find(url, A):global Listprint('正在检测图片总数，请稍等.....')t = 0i = 1s = 0while t < 1000:Url = url + str(t)try:# 这里搞了下Result = A.get(Url, timeout=7, allow_redirects=False)except BaseException:t = t + 60continueelse:result = Result.textpic_url = re.findall('"objURL":"(.*?)",', result, re.S)# 先利用正则表达式找到图片urls += len(pic_url)if len(pic_url) == 0:breakelse:List.append(pic_url)t = t + 60return sdef recommend(url):Re = []try:html = requests.get(url, allow_redirects=False)except error.HTTPError as e:returnelse:html.encoding = 'utf-8'bsObj = BeautifulSoup(html.text, 'html.parser')div = bsObj.find('div', id='topRS')if div is not None:listA = div.findAll('a')for i in listA:if i is not None:Re.append(i.get_text())return Redef dowmloadPicture(html, keyword):global num# t =0pic_url = re.findall('"objURL":"(.*?)",', html, re.S)# 先利用正则表达式找到图片urlprint('找到关键词:' + keyword + '的图片，即将开始下载图片...')for each in pic_url:print('正在下载第' + str(num + 1) + '张图片，图片地址:' + str(each))try:if each is not None:pic = requests.get(each, timeout=7)else:continueexcept BaseException:print('错误，当前图片无法下载')continueelse:string = file + r'\\' + keyword + '_' + str(num) + '.jpg'fp = open(string, 'wb')fp.write(pic.content)fp.close()num += 1if num >= numPicture:return#############################################################################################if __name__ == '__main__':# 主函数入口# 爬虫部分，存放到 name + ‘文件'#############################################################################################if GET_PIC == 1:headers = {'Accept-Language': 'zh-CN,zh; q=0.8,zh-TW; q=0.7,zh-HK; q=0.5,en-US; q=0.3,en; q=0.2','Connection': 'keep-alive','User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Firefox/60.0','Upgrade-Insecure-Requests': '1'}A = requests.Session()A.headers = headerstm = int(input('请输入每类图片的下载数量 '))numPicture = tmline_list = []with open('./name.txt', encoding='utf-8') as file:line_list = [k.strip() for k in file.readlines()]# 用 strip()移除末尾的空格for word in line_list:url = 'https://image.baidu.com/search/flip?tn=baiduimage&ie=utf-8&word=' + word + '&pn='tot = Find(url, A)Recommend = recommend(url)# 记录相关推荐print('经过检测%s类图片共有%d张' % (word, tot))file = word + '文件'y = os.path.exists(file)if y == 1:print('该文件已存在，无需创建')else:os.mkdir(file)t = 0tmp = urlwhile t < numPicture:try:url = tmp + str(t)# result = requests.get(url, timeout=10)# 这里搞了下result = A.get(url, timeout=10, allow_redirects=False)print(url)except error.HTTPError as e:print('网络错误，请调整网络后重试')t = t + 60else:dowmloadPicture(result.text, word)t = t + 60numPicture = numPicture + tmprint('当前搜索结束，开始提取人脸')############################################################################################## 将训练集和测试集中的raw图片处理，提取出人脸图片,从file+'文件'到‘待分辨人脸'############################################################################################if GET_FACE == 1:if GET_PIC == 0:file = '韩国女艺人文件'raw_root = 'E:/Table/学习数据集/find_iu/'+ file + '/'img_raw = os.listdir(raw_root)new_path = 'E:/Table/学习数据集/find_iu/待分辨人脸/'n = 0for i in range(len(img_raw)):try:img = Image.open(raw_root + img_raw[i])except:print('a file error, continue')continueelse:img_train = find_face(raw_root + img_raw[i])if img_train == None:continueelse:img_train.save(new_path + '%d.JPG' % n)print(raw_root + img_raw[i])n += 1print('在%d张图片中，共找到%d张脸' % (len(img_raw), n))print('提取人脸结束，开始寻找IU')############################################################################################## 开始判别，从'待分辨人脸‘中找出IU存放到'IU_pic‘#############################################################################################if GET_IU == 1:data_transform = transforms.Compose([# transforms.Grayscale(num_output_channels=1),# 彩色图像转灰度图像num_output_channels默认1transforms.Resize(112),transforms.CenterCrop((112, 92)),# 中心裁剪为112*92transforms.ToTensor(),transforms.Normalize(mean=[0.5, 0.5, 0.5],std=[0.5, 0.5, 0.5])# transforms.Normalize(mean=0.5, std=0.5)])device = torch.device('cuda')model = SiameNetwork().to(device)model.load_state_dict(torch.load('parameter.pkl'))# loadmodel.eval()judge_root = 'E:/Table/学习数据集/find_iu/待分辨人脸/'img_judge = os.listdir(judge_root)new_path = 'E:/Table/学习数据集/find_iu/IU_pic/'result = []n = 0for i in range(len(img_judge)):try:img = Image.open(judge_root + img_judge[i])except:print('a file error, continue')continueelse:img = img.convert('RGB')print(judge_root + img_judge[i])input = data_transform(img)input = input.unsqueeze(0)# 这里经过转换后输出的input格式是[C,H,W],网络输入还需要增加一维批量大小B# 增加一维，输出的img格式为[1,C,H,W]input = Variable(input.cuda())output = model(input)# 将图片输入网络得到输出_, id = torch.max(output.data, 1)# 0是IU,1是其他if id.item() == 0:shutil.copy(judge_root + img_judge[i], new_path)n += 1print('/n在%d张图片中，共找到%d张IU的图片'%(len(img_judge), n))#############################################################################################

4.file_deal.py

import osraw_train_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/train/IU/'raw_train_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/train/Other/'raw_test_root_1 = 'E:/Table/学习数据集/find_iu/data/raw/test/IU/'raw_test_root_2 = 'E:/Table/学习数据集/find_iu/data/raw/test/Other/'raw_roots = [raw_train_root_1, raw_train_root_2, raw_test_root_1, raw_test_root_2]for path in raw_roots:# 获取该目录下所有文件，存入列表中fileList = os.listdir(path)n = 0for i in fileList:# 设置旧文件名（就是路径+文件名）oldname = path + os.sep + fileList[n]# os.sep添加系统分隔符# 设置新文件名newname = path + os.sep + str(n) + '.JPG'os.rename(oldname, newname)# 用os模块中的rename方法对文件改名print(oldname, '======>', newname)n += 1

5.network.py

import torchimport torch.nn as nnclass SiameNetwork(nn.Module):def __init__(self):super(SiameNetwork, self).__init__()# input: h=112, w=92self.conv1 = torch.nn.Sequential(torch.nn.Conv2d(in_channels=3,# 输入单通道out_channels=16,# 16个3*3卷积核kernel_size=3,# 卷积核尺寸stride=2,# 卷积核滑动步长, 1的话图片大小不变，2的话会大小会变为(h/2)*(w/2)padding=1),# 边缘填充大小，如果要保持原大小，kernel_size//2torch.nn.BatchNorm2d(16),# 标准化，前面卷积后有16个图层torch.nn.ReLU()# 激活函数)# output: h=56, w=46self.conv2 = torch.nn.Sequential(torch.nn.Conv2d(16, 32, 3, 2, 1),torch.nn.BatchNorm2d(32),torch.nn.ReLU())# output: h=28, w=23self.conv3 = torch.nn.Sequential(torch.nn.Conv2d(32, 64, 3, 2, 1),torch.nn.BatchNorm2d(64),torch.nn.ReLU())# output: h=14, w=12self.conv4 = torch.nn.Sequential(torch.nn.Conv2d(64, 64, 2, 2, 0),torch.nn.BatchNorm2d(64),torch.nn.ReLU())# output: h=7, w=6self.mlp1 = torch.nn.Linear(7 * 6 * 64, 100)# 需要计算conv4的输出尺寸，每次卷积的输出尺寸(size - kernal + 2*padding)/stride + 1self.mlp2 = torch.nn.Linear(100, 10)def forward(self, x):x = self.conv1(x)x = self.conv2(x)x = self.conv3(x)x = self.conv4(x)x = self.mlp1(x.view(x.size(0), -1))# view展平x = self.mlp2(x)return x

6.utils.py

import matplotlib as mplimport matplotlib.pyplot as pltimport numpy as npimport cv2# 绘制训练、测试的损失、准确度#############################################################################################def draw_result(draw_epoch, path):show_loss = np.loadtxt('%s/train_loss.txt' % (path))# 读取txt文件，不同优化器的损失show_train_acc = np.loadtxt('%s/train_acc.txt' % (path))# 读取不同模型的准确度show_val_loss = np.loadtxt('%s/val_loss.txt' % (path))# 读取txt文件，不同优化器的损失show_val_acc = np.loadtxt('%s/val_acc.txt' % (path))# 读取不同模型的准确度mpl.rc('font',family='Times New Roman', weight='semibold', size=9)# 设置matplotlib中所有绘图风格的设置font1 = {'weight' : 'semibold', 'size' : 11}#设置文字风格fig = plt.figure(figsize = (7,5))#figsize是图片的大小`ax1 = fig.add_subplot(2, 2, 1)# ax1是子图的名字ax1.plot(draw_epoch, show_loss,color = 'red', label = u'AdaPID', linewidth =1.0)ax1.legend()#显示图例ax1.set_title('Training Loss', font1)ax1.set_xlabel(u'Epoch', font1)ax2 = fig.add_subplot(2, 2, 2)ax2.plot(draw_epoch, show_val_loss,color = 'red', label = u'Adam', linewidth =1.0)ax2.legend()#显示图例ax2.set_title('Validation Loss', font1)ax2.set_xlabel(u'Epoch', font1)ax3 = fig.add_subplot(2, 2, 3)ax3.plot(draw_epoch, show_train_acc,color = 'red', label = u'Adam', linewidth =1.0)ax3.legend()#显示图例ax3.set_title('Training Accuracy', font1)ax3.set_xlabel(u'Epoch', font1)ax4 = fig.add_subplot(2, 2, 4)ax4.plot(draw_epoch, show_val_acc,color = 'red', label = u'Adam', linewidth =1.0)ax4.legend()#显示图例ax4.set_title('Validation Accuracy', font1)ax4.set_xlabel(u'Epoch', font1)plt.subplots_adjust(left=None, bottom=None, right=None, top=None, wspace=None, hspace=0.45) # hspace为子图上下间距plt.savefig('%s/show_curve.jpg' % (path), dpi=300)############################################################################################## 用于解决cv.imread不能读取中文路径的问题#############################################################################################def cv_imread(filePath):# 核心就是下面这句，一般直接用这句就行，直接把图片转为mat数据cv_img = cv2.imdecode(np.fromfile(filePath, dtype=np.uint8), -1)# imdecode读取的是rgb，如果后续需要opencv处理的话，需要转换成bgr，转换后图片颜色会变化# cv_img=cv2.cvtColor(cv_img,cv2.COLOR_RGB2BGR)return cv_img#############################################################################################

总结
总体而言，这是一个新人的兴趣之作，但是限于GPU性能无法使用太复杂的网络，最后识别的效果不佳，若读者有兴趣，也可以去替换一下网络，改善一下数据集，尝试提升识别性能。更多相关人脸识别内容请搜索脚本之家以前的文章或继续浏览下面的相关文章,希望大家以后多多支持脚本之家！