人脸口罩检测(含运行代码+数据集)

  • 本教程目的为让开发者了解深度学习中的完整流程，这包括：
    1.数据集导入及预处理流程
    2.网络模型选择及参数设置流程
    3.模型训练及导出流程
    4.模型加载/优化并得出推断结果

项目源码以及数据集下载:
https://download.csdn.net/download/kunhe0512/85360655

  • 本教程采用了以下主要的软硬件环境：
    1.NVIDIA Xavier NX
    2.Jetpack 4.6
    3.TensorRT 8.0.1
    4.Pytorch 1.10.0
    5.Python 3.6.9
    6.Opencv 4.1.1

  • 实验内容：

    • 本教程的实验内容是利用深度学习的方法，完成口罩检测的任务。
    • 检测目标类别为：Background，face，mask，mask_weared_incorrect
    • 在实验过程中，采用了OpenImages CVS格式的数据集和SSD-mobilenet的模型。
    • 本实验利用Pytorch进行模型训练，将训练好的模型转化为ONNX格式，最后利用TensorRT进行推理
    • 更多精彩内容，请扫描下方二维码来加入NVIDIA开发者计划

开始实验

1.导入需要的工具库

#1import osimport syssys.executableimport loggingimport argparseimport itertoolsimport torchfrom torch.utils.data import DataLoader, ConcatDatasetfrom torch.optim.lr_scheduler import CosineAnnealingLR, MultiStepLRfrom vision.utils.misc import str2bool, Timer, freeze_net_layers, store_labelsfrom vision.ssd.ssd import MatchPriorfrom vision.ssd.vgg_ssd import create_vgg_ssdfrom vision.ssd.mobilenetv1_ssd import create_mobilenetv1_ssdfrom vision.ssd.mobilenetv1_ssd_lite import create_mobilenetv1_ssd_litefrom vision.ssd.mobilenet_v2_ssd_lite import create_mobilenetv2_ssd_litefrom vision.ssd.squeezenet_ssd_lite import create_squeezenet_ssd_litefrom vision.datasets.voc_dataset import VOCDatasetfrom vision.datasets.open_images import OpenImagesDatasetfrom vision.nn.multibox_loss import MultiboxLossfrom vision.ssd.config import vgg_ssd_configfrom vision.ssd.config import mobilenetv1_ssd_configfrom vision.ssd.config import squeezenet_ssd_configfrom vision.ssd.data_preprocessing import TrainAugmentation, TestTransform

2.使用GPU完成训练

#2DEVICE = torch.device("cuda:0")torch.backends.cudnn.benchmark = True

3.设定训练方法

#3def train(loader, net, criterion, optimizer, device, debug_steps=100, epoch=-1):net.train(True)running_loss = 0.0running_regression_loss = 0.0running_classification_loss = 0.0for i, data in enumerate(loader):images, boxes, labels = dataimages = images.to(device)boxes = boxes.to(device)labels = labels.to(device)optimizer.zero_grad()confidence, locations = net(images)regression_loss, classification_loss = criterion(confidence, locations, labels, boxes)# TODO CHANGE BOXESloss = regression_loss + classification_lossloss.backward()optimizer.step()running_loss += loss.item()running_regression_loss += regression_loss.item()running_classification_loss += classification_loss.item()if i and i % debug_steps == 0:avg_loss = running_loss / debug_stepsavg_reg_loss = running_regression_loss / debug_stepsavg_clf_loss = running_classification_loss / debug_stepsprint(f"Epoch: {epoch}, Step: {i}/{len(loader)}, " +f"Avg Loss: {avg_loss:.4f}, " +f"Avg Regression Loss {avg_reg_loss:.4f}, " +f"Avg Classification Loss: {avg_clf_loss:.4f}")running_loss = 0.0running_regression_loss = 0.0running_classification_loss = 0.0

4.设定测试方法

#4def test(loader, net, criterion, device):net.eval()running_loss = 0.0running_regression_loss = 0.0running_classification_loss = 0.0num = 0for _, data in enumerate(loader):images, boxes, labels = dataimages = images.to(device)boxes = boxes.to(device)labels = labels.to(device)num += 1with torch.no_grad():confidence, locations = net(images)regression_loss, classification_loss = criterion(confidence, locations, labels, boxes)loss = regression_loss + classification_lossrunning_loss += loss.item()running_regression_loss += regression_loss.item()running_classification_loss += classification_loss.item()return running_loss / num, running_regression_loss / num, running_classification_loss / num

5.设定训练参数

#5net_name = "mb1-ssd"datasets = []datasets_path = ["data/mask"]model_dir = "models/mask/" voc_or_open_images = "open_images"batch_size = 4num_epochs = 6validation_epochs = 2num_workers = 2lr = 0.01base_net_lr = 0.001extra_layers_lr = 0.01momentum=0.9weight_decay=5e-4

6.加载数据集

#6timer = Timer()create_net = create_mobilenetv1_ssdconfig = mobilenetv1_ssd_config# create data transforms for train/test/valtrain_transform = TrainAugmentation(config.image_size, config.image_mean, config.image_std)target_transform = MatchPrior(config.priors, config.center_variance,config.size_variance, 0.5)test_transform = TestTransform(config.image_size, config.image_mean, config.image_std)# load datasets (could be multiple)print("Prepare training datasets.")for dataset_path in datasets_path:if voc_or_open_images == 'voc':dataset = VOCDataset(dataset_path, transform=train_transform,target_transform=target_transform)label_file = os.path.join(model_dir, "labels.txt")store_labels(label_file, dataset.class_names)num_classes = len(dataset.class_names)elif voc_or_open_images == 'open_images':dataset = OpenImagesDataset(dataset_path,transform=train_transform, target_transform=target_transform,dataset_type="train", balance_data=False)label_file = os.path.join(model_dir, "labels.txt")store_labels(label_file, dataset.class_names)print(dataset)num_classes = len(dataset.class_names)else:raise ValueError(f"Dataset type is not supported.")datasets.append(dataset)

7.将加载好的数据集分割为训练集和验证集

#7# create training datasetprint(f"Stored labels into file {label_file}.")train_dataset = ConcatDataset(datasets)print("Train dataset size: {}".format(len(train_dataset)))train_loader = DataLoader(train_dataset, batch_size,num_workers=num_workers,shuffle=True) # create validation dataset print("Prepare Validation datasets.")if voc_or_open_images == "voc":val_dataset = VOCDataset(dataset_path, transform=test_transform,target_transform=target_transform, is_test=True)elif voc_or_open_images == 'open_images':val_dataset = OpenImagesDataset(dataset_path,transform=test_transform, target_transform=target_transform,dataset_type="test")print(val_dataset)print("Validation dataset size: {}".format(len(val_dataset)))val_loader = DataLoader(val_dataset, batch_size,num_workers = num_workers,shuffle=False)

8.创建网络模型

#8# create the networkprint("Build network.")net = create_net(num_classes)min_loss = -10000.0last_epoch = -1params = [{'params': net.base_net.parameters(), 'lr': base_net_lr},{'params': itertools.chain(net.source_layer_add_ons.parameters(),net.extras.parameters()), 'lr': extra_layers_lr},{'params': itertools.chain(net.regression_headers.parameters(),net.classification_headers.parameters())}]

9.定义是否使用预训练模型或者

  • 我们这里设计了三种模式:
    1.重头开始训练，只需将你的模型路径赋值给base_net： base_net = “path/to/the/basic/model”
    2.使用之前训练一半中间断开没训练完的模型继续训练，只需将模型路径赋值给resume：resume = “path/to/the/resume/model”
    3.利用我们已经准好的预训练模型，只需将模型路径赋值给pretrained_ssd: pretrained_ssd = “path/to/the/pretrained_ssd/model”
  • 如果不太明白想选择什么模型，可以将resume，base_net和pretrained_ssd都赋值None，将会自动从头开始训练
#9# load a previous model checkpoint (if requested)timer.start("Load Model")resume=Nonebase_net = Nonepretrained_ssd = "models/face-mask-pretrain-model.pth"if resume:print(f"Resume from the model {resume}")net.load(resume)elif base_net:print(f"Init from base net {base_net}")net.init_from_base_net(base_net)elif pretrained_ssd:print(f"Init from pretrained ssd {pretrained_ssd}")net.init_from_pretrained_ssd(pretrained_ssd)print(f'Took {timer.end("Load Model"):.2f} seconds to load the model.')

10.开始训练模型

#10# move the model to GPUnet.to(DEVICE)# define loss function and optimizercriterion = MultiboxLoss(config.priors, iou_threshold=0.5, neg_pos_ratio=3,center_variance=0.1, size_variance=0.2, device=DEVICE)optimizer = torch.optim.SGD(params, lr=lr, momentum=0.9, weight_decay=weight_decay)print(f"Learning rate: {lr}, Base net learning rate: {base_net_lr}, "+ f"Extra Layers learning rate: {extra_layers_lr}.")# set learning rate policyprint("Uses CosineAnnealingLR scheduler.")scheduler = CosineAnnealingLR(optimizer, 100, last_epoch=last_epoch)# train for the desired number of epochsprint(f"Start training from epoch {last_epoch + 1}.")for epoch in range(last_epoch + 1, num_epochs):scheduler.step()train(train_loader, net, criterion, optimizer,device=DEVICE, debug_steps=10, epoch=epoch)if epoch % validation_epochs == 0 or epoch == num_epochs - 1:val_loss, val_regression_loss, val_classification_loss = test(val_loader, net, criterion, DEVICE)print(f"Epoch: {epoch}, " +f"Validation Loss: {val_loss:.4f}, " +f"Validation Regression Loss {val_regression_loss:.4f}, " +f"Validation Classification Loss: {val_classification_loss:.4f}")model_path = os.path.join(model_dir, f"{net_name}-Epoch-{epoch}-Loss-{val_loss}.pth")net.save(model_path)print(f"Saved model {model_path}")print("Task done, exiting program.")

11.将训练好的模型转化成ONNX格式

#11!python3 onnx_export.py --model-dir=models/mask

12.将转化好的ONNX格式利用TensorRT进行优化，生成TensorRT推理引擎

这里注意,需要安装Onnx2TensorRT

#12!onnx2trt models/mask/ssd-mobilenet.onnx -o models/TRT_ssd_mobilenet_v1_face2.bin

13.加载引擎推理时所需要的工具库

#13import sysimport timeimport argparseimport cv2import pycuda.autoinit import numpy as npfrom utils.ssd_classes import get_cls_dictfrom utils.camera import add_camera_args, Camerafrom utils.display import open_window, set_display, show_fpsfrom utils.visualization import BBoxVisualizationimport ctypesimport tensorrt as trtimport pycuda.driver as cuda

14.设计引擎输入输出的预处理方法和后处理方法

#14def do_nms(det, boxes, confs, clss):drop = Falseif len(boxes)  0.6 and not drop:if det[4] > confs[i]:boxes[i] = ((det[0],det[1],det[2],det[3]))confs[i] = det[4]clss[i] = det[5]drop = Trueif not drop:boxes.append((det[0],det[1],det[2],det[3]))confs.append(det[4])clss.append(det[5])return boxes, confs, clssdef _preprocess_trt(img, shape=(300, 300)):"""Preprocess an image before TRT SSD inferencing."""img = cv2.resize(img, shape)img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)img = img.transpose((2, 0, 1)).astype(np.float32)img *= (2.0/255.0)img -= 1.0return imgdef _postprocess_trt(img, output, conf_th, output_layout):"""Postprocess TRT SSD output."""img_h, img_w, _ = img.shapeboxes, confs, clss, results = [], [], [],[]#print(((len(output[1]))/4+1))#print("len(outputs[0]): "+str(len(output[0]))+" len(outputs[1]): "+str(len(output[1])))for n in range(0, int((len(output[1]))/4)):maxScore = -1000.0000maxClass = 0for m in range(0, 4):score = output[0][n*4+m]#print(score)if score < conf_th:continueif m  maxScore):maxScore = scoremaxClass = m#if(maxClass < 0):#continueindex = int(n)if maxScore < conf_th:continue#print(str(output[1][n*4+0])+" "+str(output[1][n*4+1])+" "+str(output[1][n*4+2])+" "+str(output[1][n*4+3]))x1 = int(output[1][n*4+0] * img_w)y1 = int(output[1][n*4+1] * img_h)x2 = int(output[1][n*4+2] * img_w)y2 = int(output[1][n*4+3] * img_h)det = [x1,y1,x2,y2,maxScore,maxClass,n]boxes, confs, clss = do_nms(det, boxes, confs, clss)return boxes, confs, clss

15.定义SSD-mobilenet v1模型的推理引擎的加载

  • 当我们已经优化好了引擎的时候，我们可以将优化好的引擎以文件的形式写到硬盘上，我们称之为序列化文件（serialized file）或PLAN文件
  • 我们下次想直接使用优化好的引擎的时候，我们可以通过读取硬盘上的序列化文件，并利用 deserialize_cuda_engine() 方法进行反序列化，生成可执行的引擎
  • 利用序列化文件生成可执行引擎可以为我们节省大量的时间
  • 不同平台（软件或硬件平台）上生成的引擎的序列化文件不能直接通用，相同平台（软件且硬件平台）或同一台设备上生成的引擎序列化文件可以直接用
#15class TrtSSD(object):"""TrtSSD class encapsulates things needed to run TRT SSD."""#加载自定义组建，这里如果TensorRT版本小于7.0需要额外生成flattenconcat的自定义组件库def _load_plugins(self):trt.init_libnvinfer_plugins(self.trt_logger, '')#加载通过Transfer Learning Toolkit生成的推理引擎def _load_engine(self):TRTbin = 'models/TRT_%s.bin' % self.modelwith open(TRTbin, 'rb') as f, trt.Runtime(self.trt_logger) as runtime:return runtime.deserialize_cuda_engine(f.read())#通过加载的引擎，生成可执行的上下文def _create_context(self):for binding in self.engine:size = trt.volume(self.engine.get_binding_shape(binding)) * \ self.engine.max_batch_size##注意：这里的host_mem需要时用pagelocked memory，以免内存被释放host_mem = cuda.pagelocked_empty(size, np.float32)cuda_mem = cuda.mem_alloc(host_mem.nbytes)self.bindings.append(int(cuda_mem))if self.engine.binding_is_input(binding):self.host_inputs.append(host_mem)self.cuda_inputs.append(cuda_mem)else:self.host_outputs.append(host_mem)self.cuda_outputs.append(cuda_mem)return self.engine.create_execution_context()#初始化引擎def __init__(self, model, input_shape, output_layout=7):"""Initialize TensorRT plugins, engine and conetxt."""self.model = modelself.input_shape = input_shapeself.output_layout = output_layoutself.trt_logger = trt.Logger(trt.Logger.INFO)self._load_plugins()self.engine = self._load_engine()self.host_inputs = []self.cuda_inputs = []self.host_outputs = []self.cuda_outputs = []self.bindings = []self.stream = cuda.Stream()self.context = self._create_context()#释放引擎，释放GPU显存，释放CUDA流def __del__(self):"""Free CUDA memories."""del self.streamdel self.cuda_outputsdel self.cuda_inputs#利用生成的可执行上下文执行推理def detect(self, img, conf_th=0.3):"""Detect objects in the input image."""img_resized = _preprocess_trt(img, self.input_shape)np.copyto(self.host_inputs[0], img_resized.ravel())#将处理好的图片从CPU内存中复制到GPU显存cuda.memcpy_htod_async(self.cuda_inputs[0], self.host_inputs[0], self.stream)#开始执行推理任务self.context.execute_async(batch_size=1,bindings=self.bindings,stream_handle=self.stream.handle)#将推理结果输出从GPU显存复制到CPU内存cuda.memcpy_dtoh_async(self.host_outputs[1], self.cuda_outputs[1], self.stream)cuda.memcpy_dtoh_async(self.host_outputs[0], self.cuda_outputs[0], self.stream)self.stream.synchronize()output = self.host_outputs#print("len(outputs[0]): "+str(len(self.host_outputs[0]))+" len(outputs[1]): "+str(len(self.host_outputs[1])))#for x in self.host_outputs[0]:#print(str(x),end=' ')#for x in self.host_outputs[1]:#print(str(x),end=' ')return _postprocess_trt(img, output, conf_th, self.output_layout)

16.设置模型库

  • 1.这里定义了多个模型库，我们选用的是人脸口罩检测，也就是最后一个ssd_mobilenet_v1_face2
  • 2.这里还定义了我们模型的输入（300,300）
#16INPUT_HW = (300, 300)SUPPORTED_MODELS = ['ssd_mobilenet_v1_coco','ssd_mobilenet_v1_egohands','ssd_mobilenet_v2_coco','ssd_mobilenet_v2_egohands','ssd_mobilenet_v2_face','ssd_resnet18_5th','ssd_mobilenet_v1_face2','ssd_mobilenet_v1_fruit']

17.开始定义方法来读取数据并将输出可视化的画到图像上

  • detect_one()方法是检测单张图片，detect_video()方法是检测视频
  • 注意：这里打印的fps值是包括将图像写到结果视频中的时间，如果取消将视频写到结果视频的功能，速度会有大幅度提升
#17-1def detect_video(video, trt_ssd, conf_th, vis,result_file_name):full_scrn = Falsefps = 0.0tic = time.time()frame_width = int(video.get(cv2.CAP_PROP_FRAME_WIDTH))frame_height = int(video.get(cv2.CAP_PROP_FRAME_HEIGHT))fps = video.get(cv2.CAP_PROP_FPS)#print(str(frame_width)+str(frame_height))##定义输入编码fourcc = cv2.VideoWriter_fourcc('M', 'P', '4', 'V')videoWriter = cv2.VideoWriter('result.AVI', fourcc, fps, (frame_width,frame_height))##开始循环检测，并将结果写到result.mp4中while True:ret,img = video.read()if img is not None:boxes, confs, clss = trt_ssd.detect(img, conf_th)#print("boxes,confs,clss: "+ str(boxes)+" "+ str(confs)+" "+str(clss))img = vis.draw_bboxes(img, boxes, confs, clss)videoWriter.write(img)toc = time.time()curr_fps = 1.0 / (toc - tic)fps = curr_fps if fps == 0.0 else (fps*0.95 + curr_fps*0.05)tic = tocprint("\rfps: "+str(fps),end="")else:break#17-2def detect_one(img, trt_ssd, conf_th, vis):full_scrn = Falsetic = time.clock()##开始检测，并将结果写到result.jpg中boxes, confs, clss = trt_ssd.detect(img, conf_th)toc = time.clock()curr_fps = (toc - tic)#print("boxes: "+str(boxes))#print("clss: "+str(clss))#print("confs: "+str(confs))img = vis.draw_bboxes(img, boxes, confs, clss)cv2.imwrite("result.jpg",img)print("time: "+str(curr_fps)+"(sec)")

18.定义main()函数，检测单张图片**

  • 您可以自行上传图像到当前文件夹，并将filename请改成您要测试的图片的名字
  • face指的是没有戴口罩的人脸，face_mask指的是带了口罩的人脸，mask_weared_incorrect指的是带了口罩但是带的不规范的人脸
#18-1def main_one():filename = "mask.jpg"result_file_name = str(filename)img = cv2.imread(filename)cls_dict = get_cls_dict("ssd_mobilenet_v1_face2".split('_')[-1])model_name ="ssd_mobilenet_v1_face2"trt_ssd = TrtSSD(model_name, INPUT_HW)vis = BBoxVisualization(cls_dict)print("start detection!")detect_one(img, trt_ssd, conf_th=0.5, vis=vis)cv2.destroyAllWindows()print("finish!")
#18-2from IPython.display import Imagemain_one()Image("result.jpg")

19.定义main()函数，检测视频

  • 您可以自行上传视频到当前文件夹，并将filename请改成您要测试的视频的名字
  • 检测视频部分由于要将检测的结果写到硬盘上，所以时间会加倍，如果要得到和单张检测相似的数据，可以将读写的语句注释掉
  • face指的是没有戴口罩的人脸，face_mask指的是带了口罩的人脸，mask_weared_incorrect指的是带了口罩但是带的不规范的人脸）
#19-1def main_loop(): filename = "face_mask_test_video.mp4"result_file_name = str(filename)video = cv2.VideoCapture(filename)cls_dict = get_cls_dict("ssd_mobilenet_v1_face2".split('_')[-1])model_name ="ssd_mobilenet_v1_face2"trt_ssd = TrtSSD(model_name, INPUT_HW)vis = BBoxVisualization(cls_dict)print("start detection!")detect_video(video, trt_ssd, conf_th=0.8, vis=vis, result_file_name=result_file_name)video.release()cv2.destroyAllWindows()print("\nfinish!")
#19-2main_loop()

20.将生成的视频转码，以便能够在Jupyter Notebook中查看

  • 这里采用的是利用GPU加速的转码技术，将输出的视频转换到MP4格式，比单纯使用CPU进行转码的速度有大幅度提升
#20!rm result-ffmpeg4.mp4!ffmpeg -i result.AVI -vcodec libx264 -f mp4 result-ffmpeg4.mp4

21.查看结果视频

#21from IPython.display import VideoVideo("result-ffmpeg4.mp4")