def __init__(self, is_local, arch="iresnet100", weights_base_path=None):
self.__device = torch.device(
"cuda:0" if torch.cuda.is_available() else "cpu")
if is_local:
# load weights from locally
self.__embedder = iresnet_local(weights_base_path, arch)
else:
# load weights from remote urls
if arch == "iresnet34":
self.__embedder = insightface.iresnet34(pretrained=True)
elif arch == "iresnet50":
self.__embedder = insightface.iresnet50(pretrained=True)
elif arch == "iresnet100":
self.__embedder = insightface.iresnet100(pretrained=True)
else:
raise ValueError("Invalid arch type for embedder")
self.__embedder.to(self.__device)
self.__embedder.eval()
# check if model on GPU
# print(next(embedder.parameters()).is_cuda)
mean = [0.5] * 3
std = [0.5 * 256 / 255] * 3
self.__preprocess_basic = transforms.Compose([
transforms.ToPILImage(),
transforms.Resize(112),
transforms.CenterCrop(112),
transforms.ToTensor(),
])
self.__normalize = transforms.Normalize(mean, std)
def get_models(mtcnn_w_path='mtcnn/weights'):
# First we create pnet, rnet, onet, and load weights from caffe model.
pnet, rnet, onet = mtcnn.get_net_caffe(mtcnn_w_path)
# Then we create a detector
detector = mtcnn.FaceDetector(pnet, rnet, onet, device='cpu')
embedder = insightface.iresnet100(pretrained=True)
embedder.eval()
return detector, embedder
def __init__(self):
embedder = insightface.iresnet100(pretrained=True)
embedder.eval()
self.model = embedder
# ToDo: move to CUDA
mean = [0.5] * 3
std = [0.5 * 256 / 255] * 3
self.preprocess = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize(mean, std)])
def __init__(self, model_type, cfg):
self.cfg = cfg
if model_type == 'iresnet34':
self.model = insightface.iresnet34(pretrained=True).to(
self.cfg.DEVICE)
elif model_type == 'iresnet50':
self.model = insightface.iresnet50(pretrained=False).to(
self.cfg.DEVICE)
elif model_type == 'iresnet100':
self.model = insightface.iresnet100(pretrained=False).to(
self.cfg.DEVICE)
self.model.eval()
self.transform = self.transform()
def main(img):
# detect faces
pnet, rnet, onet = mtcnn.get_net_caffe('output/converted')
detector = mtcnn.FaceDetector(pnet, rnet, onet, device='cuda:0')
#img = '../FaceDetector/tests/asset/images/office5.jpg'
img = imread(img)
boxes, landmarks = detector.detect(img)
img = torch.tensor(img.astype(np.float32),
device=torch.device("cuda:0")).permute(2, 0, 1)
# embed faces
embedder = insightface.iresnet100(pretrained=True)
embedder.eval()
mean = [127.5] * 3 #[0.5] * 3
std = [128.] * 3 #[0.5 * 256 / 255] * 3
preprocess = transforms.Compose([transforms.Normalize(mean, std)])
landmarks = landmarks.float()
boxcpu = boxes.cpu().numpy()
for f0 in range(boxcpu.shape[0]):
angle = torch.atan2(landmarks[f0][1, 1] - landmarks[f0][0, 1], \
landmarks[f0][1, 0] - landmarks[f0][0, 0])# + np.pi/2
local_patch = img[:, boxes[f0][1]:boxes[f0][3],
boxes[f0][0]:boxes[f0][2]].unsqueeze(0)
local_patch = rot_img(local_patch, angle, \
[1./local_patch.shape[2]*\
(-0.5*(boxes[f0][2]+boxes[f0][0])+landmarks[f0][2, 0]), \
1./local_patch.shape[3]*\
(-0.5*(boxes[f0][3]+boxes[f0][1])+landmarks[f0][2, 1])], \
dtype=torch.cuda.FloatTensor)
local_patch = local_patch.squeeze(0)
tensor_face = F.interpolate(local_patch, size=112)
tensor_face = tensor_face.permute(0, 2, 1)
tensor_face = F.interpolate(tensor_face, size=112)
tensor_face = tensor_face.permute(0, 2, 1)
tensor_face = preprocess(tensor_face.cpu())
with torch.no_grad():
features = embedder(tensor_face.unsqueeze(0))[0].numpy()
print(features)
def main():
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
# model920 = model_920().to(device)
model920_facenet_criterion = nn.MSELoss()
# model921 = model_921()
# model921_facenet_criterion = nn.MSELoss()
InceptionResnet_model_1 = InceptionResnetV1(
pretrained='vggface2').eval().to(device)
print('load InceptionResnet-vggface2.pt successfully')
InceptionResnet_model_2 = InceptionResnetV1(
pretrained='casia-webface').eval().to(device)
print('load InceptionResnet-casia-webface.pt successfully')
IR_50_model_1 = IR_50([112, 112])
IR_50_model_1.load_state_dict(
torch.load(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/backbone_ir50_asia.pth'
))
IR_50_model_1.eval().to(device)
print('load IR_50 successfully')
IR_152_model_1 = IR_152([112, 112])
IR_152_model_1.load_state_dict(
torch.load(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/Backbone_IR_152_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'
))
IR_152_model_1.eval().to(device)
print('load IR_152 successfully')
# IR_152_model_2 = IR_152([112, 112])
# IR_152_model_2.load_state_dict(
# torch.load(
# '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/Head_ArcFace_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'))
# IR_152_model_2.eval().to(device)
# print('load IR_152_ArcFace successfully')
import insightface
# Insightface_iresent100 = insightface.iresnet100(pretrained=True)
# Insightface_iresent100.eval().to(device)
# print('load Insightface_iresent100 successfully')
Insightface_iresnet34 = insightface.iresnet34(pretrained=True)
Insightface_iresnet34.eval().to(device)
print('load Insightface_iresnet34 successfully')
Insightface_iresnet50 = insightface.iresnet50(pretrained=True)
Insightface_iresnet50.eval().to(device)
print('load Insightface_iresnet50 successfully')
Insightface_iresnet100 = insightface.iresnet100(pretrained=True)
Insightface_iresnet100.eval().to(device)
print('load Insightface_iresnet100 successfully')
criterion = nn.MSELoss()
# cpu
# collect all images to attack
paths = []
picpath = '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/images'
for root, dirs, files in os.walk(picpath):
for f in files:
paths.append(os.path.join(root, f))
random.shuffle(paths)
# paras
eps = 1
steps = 75
output_path = './output_img'
momentum = 1.0
for path in tqdm(paths):
start = time.time()
print('processing ' + path + ' ===============>')
image = Image.open(path)
# define paras
# in_tensor is origin tensor of image
# in_variable changes with gradient
in_tensor = img2tensor(np.array(image))
# print(in_tensor.shape)
in_variable = in_tensor.detach().to(device)
in_tensor = in_tensor.squeeze().to(device)
adv = None
# in_tensor= img2tensor_224(image)
# # print(in_tensor.shape)
# in_variable = in_tensor.to(device)
# in_tensor = in_tensor.squeeze().to(device)
# adv = None
#
# # origin feature
# origin_model920 = model920(in_variable).to(device)
# origin_model921 = model921(in_variable)
origin_InceptionResnet_model_1 = InceptionResnet_model_1(in_variable)
origin_InceptionResnet_model_2 = InceptionResnet_model_2(in_variable)
origin_IR_50_model_1 = IR_50_model_1(in_variable)
origin_IR_152_model_1 = IR_152_model_1(in_variable)
# origin_IR_152_model_2 = IR_152_model_2(in_variable)
origin_Insightface_iresent34 = Insightface_iresnet34(in_variable)
origin_Insightface_iresent50 = Insightface_iresnet50(in_variable)
origin_Insightface_iresent100 = Insightface_iresnet100(in_variable)
# 1. untarget attack -> random noise
# 2. target attack -> x = alpha * target + (1 - alpha) * x
perturbation = torch.Tensor(3, 112, 112).uniform_(-0.1, 0.1).to(device)
in_variable = in_variable + perturbation
in_variable.data.clamp_(-1.0, 1.0)
in_variable.requires_grad = True
g_noise = 0.0
# sum gradient
for i in range(steps):
# print('step: ' + str(i))
# in_variable = in_variable.to(device)
# out_model920 = model920(in_variable)
# out_model921 = model921(in_variable)
out_InceptionResnet_model_1 = InceptionResnet_model_1(in_variable)
out_InceptionResnet_model_2 = InceptionResnet_model_2(in_variable)
out_IR_50_model_1 = IR_50_model_1(in_variable)
out_IR_152_model_1 = IR_152_model_1(in_variable)
# out_IR_152_model_2 = IR_152_model_2(in_variable)
out_Insightface_iresent34 = Insightface_iresnet34(in_variable)
out_Insightface_iresent50 = Insightface_iresnet50(in_variable)
out_Insightface_iresent100 = Insightface_iresnet100(in_variable)
loss = criterion(origin_InceptionResnet_model_1, out_InceptionResnet_model_1) + \
criterion(origin_InceptionResnet_model_2, out_InceptionResnet_model_2) + \
criterion(origin_IR_50_model_1, out_IR_50_model_1) + \
criterion(origin_IR_152_model_1, out_IR_152_model_1) + \
criterion(origin_Insightface_iresent34, out_Insightface_iresent34) + \
criterion(origin_Insightface_iresent50, out_Insightface_iresent50) +\
criterion(origin_Insightface_iresent100, out_Insightface_iresent100)
# print('loss : %f' % loss)
# compute gradients
loss.backward(retain_graph=True)
g_noise = momentum * g_noise + (in_variable.grad /
in_variable.grad.data.norm(1))
g_noise = g_noise / g_noise.data.norm(1)
if i % 2 == 0:
kernel = gkern(3, 2).astype(np.float32)
gaussian_blur1 = GaussianBlur(kernel).to(device)
g_noise = gaussian_blur1(g_noise)
g_noise = torch.clamp(g_noise, -0.1, 0.1)
else:
addition = TVLoss()
g_noise = addition(g_noise)
in_variable.data = in_variable.data + (
(eps / 255.) * torch.sign(g_noise)
) # * torch.from_numpy(mat).unsqueeze(0).float()
in_variable.grad.data.zero_() # unnecessary
# deprocess image
adv = in_variable.data.cpu().numpy()[0] # (3, 112, 112)
perturbation = (adv - in_tensor.cpu().numpy())
adv = adv * 128.0 + 127.0
adv = adv.swapaxes(0, 1).swapaxes(1, 2)
adv = adv[..., ::-1]
adv = np.clip(adv, 0, 255).astype(np.uint8)
sample_dir = '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/main_3_output-8-28/'
if not os.path.exists(sample_dir):
os.makedirs(sample_dir)
advimg = sample_dir + path.split('/')[-1].split('.')[-2] + '.jpg'
cv2.imwrite(advimg, adv)
print("save path is " + advimg)
print('cost time is %.2f s ' % (time.time() - start))
def main():
sample_dir = './test_DI-2-FGSM-3/'
if not os.path.exists(sample_dir):
os.makedirs(sample_dir)
InceptionResnet_model_1 = InceptionResnetV1(
pretrained='vggface2').eval().to(device_0)
print('load InceptionResnet-vggface2.pt successfully')
# InceptionResnet_model_2 = InceptionResnetV1(pretrained='casia-webface').eval().to(device_0)
# print('load InceptionResnet-casia-webface.pt successfully')
IR_50_model_1 = IR_50([112, 112])
IR_50_model_1.load_state_dict(
torch.load(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/backbone_ir50_asia.pth'
))
IR_50_model_1.eval().to(device_0)
print('load IR_50 successfully')
# IR_152_model_1 = IR_152([112, 112])
# IR_152_model_1.load_state_dict(
# torch.load(
# '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/Backbone_IR_152_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'))
# IR_152_model_1.eval().to(device_0)
# print('load IR_152 successfully')
# IR_SE_50 = Backbone(50,mode='ir_se').eval().to(device_1)
# print('load IR_SE_50 successfully')
# mobileFaceNet = MobileFaceNet(512).eval().to(device_0)
# print('load mobileFaceNet successfully')
Insightface_iresnet34 = insightface.iresnet34(pretrained=True)
Insightface_iresnet34.eval().to(device_1)
print('load Insightface_iresnet34 successfully')
Insightface_iresnet50 = insightface.iresnet50(pretrained=True)
Insightface_iresnet50.eval().to(device_1)
print('load Insightface_iresnet50 successfully')
Insightface_iresnet100 = insightface.iresnet100(pretrained=True)
Insightface_iresnet100.eval().to(device_1)
print('load Insightface_iresnet100 successfully')
# ##########################vgg16
# from Face_recognition.vgg16.vgg16 import CenterLossModel,loadCheckpoint
# vgg16_checkpoint=loadCheckpoint('/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/vgg16/model');
#
# VGG16 = CenterLossModel(embedding_size=512,num_classes=712,checkpoint=vgg16_checkpoint).eval().to(device_1)
# print('load VGG16 successfully')
arc_face_ir_se_50 = Arcface()
arc_face_ir_se_50.eval()
arc_face_ir_se_50.to(device_0)
models = []
models.append(InceptionResnet_model_1)
models.append(IR_50_model_1)
models.append(Insightface_iresnet34)
models.append(Insightface_iresnet50)
models.append(Insightface_iresnet100)
models.append(arc_face_ir_se_50)
criterion = nn.MSELoss()
# cpu
# collect all images to attack
paths = []
picpath = '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/images'
for root, dirs, files in os.walk(picpath):
for f in files:
paths.append(os.path.join(root, f))
random.shuffle(paths)
# paras
eps = 1
steps = 50
output_path = './output_img'
momentum = 0.3
alpha = 0.35
beta = 0.3
gamma = 0.1
#####cal mean feature face
print('cal mean feature face #########################')
# ##########
# cal mean feature face on only 712 images
# mean_face_1 = torch.zeros(1,3,112,112).detach()
# for path in tqdm(paths):
# image = Image.open(path)
# in_tensor_1 = img2tensor(np.array(image))
# mean_face_1 += in_tensor_1
# mean_face_1 = mean_face_1 / 712
# ##########
# with torch.no_grad():
# mean_face = torch.zeros(512).detach()
# for path in tqdm(paths):
# start = time.time()
# print('cal mean face ' + path + ' ===============>')
# image = Image.open(path)
#
#
# # define paras
# # in_tensor is origin tensor of image
# # in_variable changes with gradient
#
# in_tensor_1 = img2tensor(np.array(image))
# # print(in_tensor.shape)
# this_feature_face = None
#
#
# # # origin feature
#
# _origin_InceptionResnet_model_1 = InceptionResnet_model_1(in_tensor_1.to(device_0)).cpu()
# ####################
# # _origin_InceptionResnet_model_2 = InceptionResnet_model_2(in_tensor_1.to(device_0)).cpu()
# ######################
# # _origin_IR_50_model_1 = IR_50_model_1(in_tensor_1.to(device_0)).cpu()
# ###########################
# # _origin_IR_152_model_1 = IR_152_model_1(in_tensor_1.to(device_0)).cpu()
# # _origin_IR_SE_50 = IR_SE_50(in_tensor_1.to(device_1)).cpu()
# # _origin_mobileFaceNet = mobileFaceNet(in_tensor_1.to(device_0)).cpu()
# #############################
#
# _origin_Insightface_iresent34 = Insightface_iresnet34(in_tensor_1.to(device_1)).cpu()
#
#
# _origin_Insightface_iresent50 = Insightface_iresnet50(in_tensor_1.to(device_1)).cpu()
#
#
# _origin_Insightface_iresent100 = Insightface_iresnet100(in_tensor_1.to(device_1)).cpu()
#
#
# _origin_arcface = arc_face_ir_se_50(in_tensor_1.to(device_0)).cpu()
#
# ########################
# # _origin_VGG16 = VGG16.forward_GetFeature(in_tensor_1.to(device_1)).cpu()
# ########################
#
# this_feature_face = _origin_InceptionResnet_model_1 + \
# _origin_Insightface_iresent34 + \
# _origin_Insightface_iresent50 + \
# _origin_Insightface_iresent100 + \
# _origin_arcface
#
# # this_feature_face = _origin_InceptionResnet_model_1 + \
# # _origin_InceptionResnet_model_2 +\
# # _origin_IR_50_model_1 + \
# # _origin_IR_152_model_1 +\
# # _origin_IR_SE_50 +\
# # _origin_mobileFaceNet +\
# # _origin_Insightface_iresent34 + \
# # _origin_Insightface_iresent50 + \
# # _origin_Insightface_iresent100 + \
# # _origin_arcface +\
# # _origin_VGG16
#
#
# this_feature_face = this_feature_face / 5.
# mean_face = mean_face + this_feature_face
#
# # del _origin_InceptionResnet_model_1
# # del _origin_InceptionResnet_model_2
# # del _origin_IR_50_model_1
# # del _origin_IR_152_model_1
# # del _origin_IR_SE_50
# # del _origin_mobileFaceNet
# # del _origin_Insightface_iresent34
# # del _origin_Insightface_iresent50
# # del _origin_Insightface_iresent100
# # del _origin_VGG16
# # del _origin_arcface
# # del this_feature_face
# # del in_tensor_1
#
# del _origin_InceptionResnet_model_1
# # del _origin_IR_50_model_1
# del _origin_Insightface_iresent34
# del _origin_Insightface_iresent50
# del _origin_Insightface_iresent100
# del _origin_arcface
# del this_feature_face
# del in_tensor_1
#
# mean_face = mean_face / 712.
mean_face = cal_mean_face_by_extend_dataset(models)
print('finish cal mean face...')
############################
print('######attack...##################')
from mydataset import CustomDataset
custom_dataset = CustomDataset()
train_loader = torch.utils.data.DataLoader(dataset=custom_dataset,
batch_size=6,
shuffle=True)
count = 0
progressRate = 0.0
for i, (x, path) in enumerate(train_loader):
start = time.time()
print('processing ' + str(progressRate) + ' ===============>')
in_tensor = x
origin_variable = in_tensor.detach()
origin_variable = origin_variable
in_variable = in_tensor.detach()
in_variable = in_variable
##########
in_variable_max = in_tensor.detach()
in_variable_min = in_tensor.detach()
in_variable_max = in_variable + 0.1
in_variable_min = in_variable - 0.1
##########
in_tensor = in_tensor.squeeze()
in_tensor = in_tensor
adv = None
perturbation = torch.Tensor(x.size(0), 3, 112,
112).uniform_(-0.05, 0.05) ###这里测试
perturbation = perturbation
in_variable += perturbation
in_variable.data.clamp_(-1.0, 1.0)
in_variable.requires_grad = True
g_noise = torch.zeros_like(in_variable)
g_noise = g_noise
origin_InceptionResnet_model_1 = InceptionResnet_model_1(
origin_variable.to(device_0)).cpu()
# origin_InceptionResnet_model_2 = InceptionResnet_model_2(origin_variable.to(device_0)).cpu()
origin_IR_50_model_1 = IR_50_model_1(
origin_variable.to(device_0)).cpu()
# origin_IR_152_model_1 = IR_152_model_1(origin_variable.to(device_0)).cpu()
# origin_IR_SE_50 = IR_SE_50(origin_variable.to(device_1)).cpu()
# origin_mobileFaceNet = mobileFaceNet(origin_variable.to(device_0)).cpu()
origin_Insightface_iresent34 = Insightface_iresnet34(
origin_variable.to(device_1)).cpu()
origin_Insightface_iresent50 = Insightface_iresnet50(
origin_variable.to(device_1)).cpu()
origin_Insightface_iresent100 = Insightface_iresnet100(
origin_variable.to(device_1)).cpu()
origin_arcface = arc_face_ir_se_50(origin_variable.to(device_0)).cpu()
# origin_VGG16 = VGG16.forward_GetFeature(origin_variable.to(device_1)).cpu()
# origin_average_out = (origin_InceptionResnet_model_1+origin_IR_50_model_1+origin_Insightface_iresent34+\
# origin_Insightface_iresent50+origin_Insightface_iresent100+origin_arcface)/6
# origin_average_out =(origin_InceptionResnet_model_1+\
# origin_InceptionResnet_model_2+ \
# origin_IR_50_model_1+\
# origin_IR_152_model_1+\
# origin_IR_SE_50+\
# origin_mobileFaceNet+\
# origin_Insightface_iresent34+\
# origin_Insightface_iresent50 +\
# origin_Insightface_iresent100 +\
# origin_arcface +\
# origin_VGG16) /11.
origin_average_out =(origin_InceptionResnet_model_1+ \
origin_IR_50_model_1 +\
origin_Insightface_iresent34+\
origin_Insightface_iresent50 +\
origin_Insightface_iresent100 +\
origin_arcface ) /6.
# target pix mean face
# After verification, it found no use.
# target_mean_face_InceptionResnet_model_1 = InceptionResnet_model_1(mean_face_1.to(device_0)).cpu()
# target_mean_face_IR_50_model_1 = IR_50_model_1(mean_face_1.to(device_0)).cpu()
# target_mean_face_Insightface_iresent34 = Insightface_iresnet34(mean_face_1.to(device_1)).cpu()
# target_mean_face_Insightface_iresent50 = Insightface_iresnet50(mean_face_1.to(device_1)).cpu()
# target_mean_faceInsightface_iresent100 = Insightface_iresnet100(mean_face_1.to(device_1)).cpu()
# target_mean_face_arcface = arc_face_ir_se_50(mean_face_1.to(device_0)).cpu()
# target_mean_face_average_out = (target_mean_face_InceptionResnet_model_1 + target_mean_face_IR_50_model_1 + target_mean_face_Insightface_iresent34 + target_mean_face_Insightface_iresent50 + target_mean_faceInsightface_iresent100 + target_mean_face_arcface)/ 6
# sum gradient
for i in range(steps):
print('step: ' + str(i))
# new_variable = input_diversity(in_variable,112,0.5)
# 通过随机的size的padding，增加input的多样性
mediate_InceptionResnet_model_1 = InceptionResnet_model_1(
in_variable.to(device_0)).cpu()
# mediate_InceptionResnet_model_2 = InceptionResnet_model_2(new_variable.to(device_0)).cpu()
mediate_IR_50_model_1 = IR_50_model_1(
in_variable.to(device_0)).cpu()
# mediate_IR_152_model_1 = IR_152_model_1(new_variable.to(device_0)).cpu()
# mediate_IR_SE_50 = IR_SE_50(new_variable.to(device_1)).cpu()
# mediate_mobileFaceNet = mobileFaceNet(new_variable.to(device_0)).cpu()
mediate_Insightface_iresent34 = Insightface_iresnet34(
in_variable.to(device_1)).cpu()
mediate_Insightface_iresent50 = Insightface_iresnet50(
in_variable.to(device_1)).cpu()
mediate_Insightface_iresent100 = Insightface_iresnet100(
in_variable.to(device_1)).cpu()
# mediate_VGG16 = VGG16.forward_GetFeature(new_variable.to(device_1)).cpu()
mediate_arcface = arc_face_ir_se_50(in_variable.to(device_0)).cpu()
# average_out = (mediate_InceptionResnet_model_1+mediate_InceptionResnet_model_2+mediate_IR_50_model_1+\
# mediate_IR_152_model_1+mediate_IR_SE_50+mediate_mobileFaceNet+mediate_Insightface_iresent34+\
# mediate_Insightface_iresent50+mediate_Insightface_iresent100+mediate_VGG16)/10
# mediate_average_out = (mediate_InceptionResnet_model_1+mediate_IR_50_model_1+mediate_Insightface_iresent34+\
# mediate_Insightface_iresent50+mediate_Insightface_iresent100+mediate_arcface)/6
# mediate_average_out = (mediate_InceptionResnet_model_1+\
# mediate_InceptionResnet_model_2+\
# mediate_IR_50_model_1+\
# mediate_IR_152_model_1+\
# mediate_IR_SE_50+\
# mediate_mobileFaceNet+\
# mediate_Insightface_iresent34+\
# mediate_Insightface_iresent50+\
# mediate_Insightface_iresent100 +\
# mediate_VGG16+\
# mediate_arcface) /11.
mediate_average_out = (mediate_InceptionResnet_model_1+ \
mediate_IR_50_model_1 +\
mediate_Insightface_iresent34+\
mediate_Insightface_iresent50+\
mediate_Insightface_iresent100 +\
mediate_arcface) /6.
# loss1 = criterion(mediate_InceptionResnet_model_1, origin_InceptionResnet_model_1) + \
# criterion(mediate_InceptionResnet_model_2, origin_InceptionResnet_model_2) + \
# criterion(mediate_IR_50_model_1, origin_IR_50_model_1) + \
# criterion(mediate_IR_152_model_1, origin_IR_152_model_1) + \
# criterion(mediate_IR_SE_50, origin_IR_SE_50) + \
# criterion(mediate_mobileFaceNet, origin_mobileFaceNet)+ \
# criterion(mediate_Insightface_iresent34, origin_Insightface_iresent34)+ \
# criterion(mediate_Insightface_iresent50, origin_Insightface_iresent50) + \
# criterion(mediate_Insightface_iresent100, origin_Insightface_iresent100) + \
# criterion(mediate_VGG16, origin_VGG16)
loss1 = criterion(mediate_average_out, origin_average_out)
# loss2 = criterion(mediate_InceptionResnet_model_1, mean_face) + \
# criterion(mediate_InceptionResnet_model_2, mean_face) + \
# criterion(mediate_IR_50_model_1, mean_face) + \
# criterion(mediate_IR_152_model_1, mean_face) +\
# criterion(mediate_IR_SE_50, mean_face) + \
# criterion(mediate_mobileFaceNet, mean_face) + \
# criterion(mediate_Insightface_iresent34, mean_face) + \
# criterion(mediate_Insightface_iresent50, mean_face) + \
# criterion(mediate_Insightface_iresent100, mean_face) + \
# criterion(mediate_VGG16, mean_face)
# loss2 = criterion(mediate_average_out, target_mean_face_average_out)
# loss3 = criterion(mediate_average_out,torch.zeros(512).detach())
loss2 = criterion(mediate_average_out, mean_face)
# loss3 = criterion(mediate_InceptionResnet_model_1,average_out)+ \
# criterion(mediate_InceptionResnet_model_2,average_out)+ \
# criterion(mediate_IR_50_model_1,average_out) + \
# criterion(mediate_IR_152_model_1,average_out) + \
# criterion(mediate_mobileFaceNet,average_out) + \
# criterion(mediate_Insightface_iresent34,average_out)+ \
# criterion(mediate_Insightface_iresent50,average_out) + \
# criterion(mediate_Insightface_iresent100,average_out)+ \
# criterion(mediate_VGG16,average_out)+ \
# criterion(mediate_IR_SE_50,average_out)
#
# loss = alpha * loss1 - beta* loss2 - gamma*loss3
loss = alpha * loss1 - beta * loss2
# print('loss : %f ' % loss,'loss1 : %f ' % loss1,'loss2 : %f ' % loss2,'loss3 : %f ' % loss3)
# compute gradients
loss.backward(retain_graph=True)
g_noise = momentum * g_noise + (in_variable.grad /
in_variable.grad.data.norm(1))
g_noise = g_noise / g_noise.data.norm(1)
g1 = g_noise
g2 = g_noise
# if i % 3 == 0 :
kernel = gkern(3, 2).astype(np.float32)
gaussian_blur1 = GaussianBlur(kernel)
gaussian_blur1
g1 = gaussian_blur1(g1)
# else:
addition = TVLoss()
addition
g2 = addition(g2)
g_noise = 0.25 * g1 + 0.75 * g2
in_variable.data = in_variable.data + (
(eps / 255.) * torch.sign(g_noise)
) # * torch.from_numpy(mat).unsqueeze(0).float()
in_variable.data = clip_by_tensor(in_variable.data,
in_variable_min.data,
in_variable_max.data)
in_variable.grad.data.zero_() # unnecessary
# del new_variable
# g_noise = in_variable.data - origin_variable
# g_noise.clamp_(-0.2, 0.2)
# in_variable.data = origin_variable + g_noise
# deprocess image
for i in range(len(in_variable.data.cpu().numpy())):
adv = in_variable.data.cpu().numpy()[i] # (3, 112, 112)
perturbation = (adv - in_tensor.cpu().numpy())
adv = adv * 128.0 + 127.0
adv = adv.swapaxes(0, 1).swapaxes(1, 2)
adv = adv[..., ::-1]
adv = np.clip(adv, 0, 255).astype(np.uint8)
# sample_dir = './target_mean_face/'
# if not os.path.exists(sample_dir):
# os.makedirs(sample_dir)
advimg = sample_dir + path[i].split('/')[-1].split(
'.')[-2] + '.jpg'
print(advimg)
cv2.imwrite(advimg, adv)
print("save path is " + advimg)
print('cost time is %.2f s ' % (time.time() - start))
count += 6
progressRate = count / 712.
# Getting an array of cropped and transformed image for each face.
faces = []
for i in range(len(landmarks)):
similar_trans_matrix = findNonreflectiveSimilarity(
np.array(landmarks[i]).astype(np.float32), REFERENCE_FACIAL_POINTS)
img_copy = img.copy()
aligned_face = cv2.warpAffine(src=np.array(img_copy),
M=similar_trans_matrix,
dsize=(112, 112))
faces.append(aligned_face)
plt.imshow(aligned_face)
plt.show()
# Defining the face embedder as a pretrained model
embedder = insightface.iresnet100(pretrained=True)
embedder.eval()
# Defining a preprocess for each face pefore passing it to the network.
mean = [0.5] * 3
std = [0.5 * 256 / 255] * 3
preprocess = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize(mean, std)])
# Getting the embedding features for each face
features_all = []
with torch.no_grad():
for face in faces:
tensor = preprocess(face)
feature_face = embedder(tensor.unsqueeze(0))
def main():
sample_dir = './target_mean_face_1/'
if not os.path.exists(sample_dir):
os.makedirs(sample_dir)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
InceptionResnet_model_1 = InceptionResnetV1(
pretrained='vggface2').eval().to(device)
print('load InceptionResnet-vggface2.pt successfully')
InceptionResnet_model_2 = InceptionResnetV1(
pretrained='casia-webface').eval().to(device)
print('load InceptionResnet-casia-webface.pt successfully')
IR_50_model_1 = IR_50([112, 112])
IR_50_model_1.load_state_dict(
torch.load(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/backbone_ir50_asia.pth'
))
IR_50_model_1.eval().to(device)
print('load IR_50 successfully')
IR_152_model_1 = IR_152([112, 112])
IR_152_model_1.load_state_dict(
torch.load(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/Backbone_IR_152_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'
))
IR_152_model_1.eval().to(device)
print('load IR_152 successfully')
IR_SE_50 = Backbone(50, mode='ir_se').eval().to(device)
print('load IR_SE_50 successfully')
mobileFaceNet = MobileFaceNet(512).eval().to(device)
print('load mobileFaceNet successfully')
# IR_152_model_2 = IR_152([112, 112])
# IR_152_model_2.load_state_dict(
# torch.load(
# '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/irse/model/Head_ArcFace_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'))
# IR_152_model_2.eval().to(device)
# print('load IR_152_ArcFace successfully')
import insightface
Insightface_iresnet34 = insightface.iresnet34(pretrained=True)
Insightface_iresnet34.eval().to(device)
print('load Insightface_iresnet34 successfully')
Insightface_iresnet50 = insightface.iresnet50(pretrained=True)
Insightface_iresnet50.eval().to(device)
print('load Insightface_iresnet50 successfully')
Insightface_iresnet100 = insightface.iresnet100(pretrained=True)
Insightface_iresnet100.eval().to(device)
print('load Insightface_iresnet100 successfully')
###########################vgg16
from Face_recognition.vgg16.vgg16 import CenterLossModel, loadCheckpoint
vgg16_checkpoint = loadCheckpoint(
'/notebooks/Workspace/tmp/pycharm_project_314/TianChi/Face_recognition/vgg16/model'
)
VGG16 = CenterLossModel(embedding_size=512,
num_classes=712,
checkpoint=vgg16_checkpoint).eval().to(device)
print('load VGG16 successfully')
# ################on swj's server
# InceptionResnet_model_1 = InceptionResnetV1(pretrained='vggface2').eval()
# print('load InceptionResnet-vggface2.pt successfully')
#
# InceptionResnet_model_2 = InceptionResnetV1(pretrained='casia-webface').eval()
# print('load InceptionResnet-casia-webface.pt successfully')
#
# IR_50_model_1 = IR_50([112, 112])
# IR_50_model_1.load_state_dict(torch.load('./face_recognition/irse/model/backbone_ir50_asia.pth'))
# IR_50_model_1.eval()
# print('load IR_50 successfully')
#
# IR_152_model_1 = IR_152([112, 112])
# IR_152_model_1.load_state_dict(torch.load(
# './face_recognition/irse/model/Backbone_IR_152_Epoch_112_Batch_2547328_Time_2019-07-13-02-59_checkpoint.pth'))
# IR_152_model_1.eval()
# print('load IR_152 successfully')
#
# IR_SE_50 = Backbone(50, mode='ir_se').eval()
# print('load IR_SE_50 successfully')
#
# mobileFaceNet = MobileFaceNet(512).eval()
# print('load mobileFaceNet successfully')
#
# Insightface_iresnet34 = insightface.iresnet34(pretrained=True)
# Insightface_iresnet34.eval()
# print('load Insightface_iresnet34 successfully')
#
# Insightface_iresnet50 = insightface.iresnet50(pretrained=True)
# Insightface_iresnet50.eval()
# print('load Insightface_iresnet50 successfully')
#
# Insightface_iresnet100 = insightface.iresnet100(pretrained=True)
# Insightface_iresnet100.eval()
# print('load Insightface_iresnet100 successfully')
#
# vgg16_checkpoint = loadCheckpoint('./face_recognition/vgg16/model')
# VGG16 = CenterLossModel(embedding_size=512, num_classes=712, checkpoint=vgg16_checkpoint).eval()
# print('load vgg16 successfully')
####load model to cuda
InceptionResnet_model_1.to(device)
InceptionResnet_model_2.to(device)
IR_50_model_1.to(device)
IR_152_model_1.to(device)
IR_SE_50.to(device)
mobileFaceNet.to(device)
Insightface_iresnet34.to(device)
Insightface_iresnet50.to(device)
Insightface_iresnet100.to(device)
VGG16.to(device)
criterion = nn.MSELoss()
# cpu
# collect all images to attack
paths = []
picpath = '/notebooks/Workspace/tmp/pycharm_project_314/TianChi/images'
for root, dirs, files in os.walk(picpath):
for f in files:
paths.append(os.path.join(root, f))
random.shuffle(paths)
# paras
eps = 1
steps = 50
output_path = './output_img'
momentum = 1.0
alpha = 0.35
beta = 0.9
gamma = 0.1
#######cal mean feature face
print('cal mean feature face #########################')
mean_face = torch.zeros(512).detach().to(device)
for path in tqdm(paths):
start = time.time()
print('cal mean face ' + path + ' ===============>')
image = Image.open(path)
# define paras
# in_tensor is origin tensor of image
# in_variable changes with gradient
in_tensor_1 = img2tensor(np.array(image))
# print(in_tensor.shape)
in_variable_1 = in_tensor_1.detach().to(device)
in_tensor_1 = in_tensor_1.squeeze().to(device)
this_feature_face = None
# # origin feature
_origin_InceptionResnet_model_1 = InceptionResnet_model_1(
in_variable_1).volatile = True
_origin_InceptionResnet_model_2 = InceptionResnet_model_2(
in_variable_1).volatile = True
_origin_IR_50_model_1 = IR_50_model_1(in_variable_1).volatile = True
_origin_IR_152_model_1 = IR_152_model_1(in_variable_1).volatile = True
_origin_IR_SE_50 = IR_SE_50(in_variable_1).volatile = True
_origin_mobileFaceNet = mobileFaceNet(in_variable_1).volatile = True
_origin_Insightface_iresent34 = Insightface_iresnet34(
in_variable_1).volatile = True
_origin_Insightface_iresent50 = Insightface_iresnet50(
in_variable_1).volatile = True
_origin_Insightface_iresent100 = Insightface_iresnet100(
in_variable_1).volatile = True
_origin_VGG16 = VGG16.forward_GetFeature(in_variable_1).volatile = True
this_feature_face = _origin_InceptionResnet_model_1*0.7 + \
_origin_InceptionResnet_model_2*0.7 + \
_origin_IR_50_model_1 *0.8+ \
_origin_IR_152_model_1 *0.8 + \
_origin_IR_SE_50 *0.7+ \
_origin_mobileFaceNet *0.7+ \
_origin_Insightface_iresent34 *0.8 + \
_origin_Insightface_iresent50 *0.9 + \
_origin_Insightface_iresent100 *0.9 + \
_origin_VGG16 *0.7
this_feature_face = this_feature_face / 10.
mean_face = mean_face + this_feature_face
del _origin_InceptionResnet_model_1
del _origin_InceptionResnet_model_2
del _origin_IR_50_model_1
del _origin_IR_152_model_1
del _origin_IR_SE_50
del _origin_mobileFaceNet
del _origin_Insightface_iresent34
del _origin_Insightface_iresent50
del _origin_Insightface_iresent100
del _origin_VGG16
del this_feature_face
del in_tensor_1
del in_variable_1
mean_face = mean_face / 712.
print('finish cal mean face...')
#############################
#####
print('######attack...##################')
for path in tqdm(paths):
start = time.time()
print('processing ' + path + ' ===============>')
image = Image.open(path)
# define paras
# in_tensor is origin tensor of image
# in_variable changes with gradient
in_tensor = img2tensor(np.array(image))
origin_variable = in_tensor.detach()
origin_variable = origin_variable.to(device)
tar_tensor = mean_face.to(device)
in_variable = in_tensor.detach()
in_variable = in_variable.to(device)
tar_variable = tar_tensor.detach()
tar_variable = tar_variable.to(device)
in_tensor = in_tensor.squeeze()
in_tensor = in_tensor.to(device)
adv = None
perturbation = torch.Tensor(3, 112, 112).uniform_(-0.05, 0.05)
perturbation = perturbation.to(device)
in_variable += perturbation
in_variable.data.clamp_(-1.0, 1.0)
in_variable.requires_grad = True
g_noise = torch.zeros_like(in_variable)
g_noise = g_noise.to(device)
origin_InceptionResnet_model_1 = InceptionResnet_model_1(
origin_variable)
origin_InceptionResnet_model_2 = InceptionResnet_model_2(
origin_variable)
origin_IR_50_model_1 = IR_50_model_1(origin_variable)
origin_IR_152_model_1 = IR_152_model_1(origin_variable)
origin_IR_SE_50 = IR_SE_50(origin_variable)
origin_mobileFaceNet = mobileFaceNet(origin_variable)
# # origin_IR_152_model_2 = IR_152_model_2(in_variable)
origin_Insightface_iresent34 = Insightface_iresnet34(origin_variable)
origin_Insightface_iresent50 = Insightface_iresnet50(origin_variable)
origin_Insightface_iresent100 = Insightface_iresnet100(origin_variable)
origin_VGG16 = VGG16.forward_GetFeature(origin_variable)
# sum gradient
for i in range(steps):
print('step: ' + str(i))
mediate_InceptionResnet_model_1 = InceptionResnet_model_1(
in_variable)
mediate_InceptionResnet_model_2 = InceptionResnet_model_2(
in_variable)
mediate_IR_50_model_1 = IR_50_model_1(in_variable)
mediate_IR_152_model_1 = IR_152_model_1(in_variable)
mediate_IR_SE_50 = IR_SE_50(in_variable)
mediate_mobileFaceNet = mobileFaceNet(in_variable)
# # origin_IR_152_model_2 = IR_152_model_2(in_variable)
mediate_Insightface_iresent34 = Insightface_iresnet34(in_variable)
mediate_Insightface_iresent50 = Insightface_iresnet50(in_variable)
mediate_Insightface_iresent100 = Insightface_iresnet100(
in_variable)
mediate_VGG16 = VGG16.forward_GetFeature(in_variable)
average_out = (mediate_InceptionResnet_model_1+mediate_InceptionResnet_model_2+mediate_IR_50_model_1+\
mediate_IR_152_model_1+mediate_IR_SE_50+mediate_mobileFaceNet+mediate_Insightface_iresent34+\
mediate_Insightface_iresent50+mediate_Insightface_iresent100+mediate_VGG16)/10
# loss1 far away from orgin image, loss2 approach target image
# loss1 = criterion(origin_InceptionResnet_model_1, mediate_InceptionResnet_model_1) + \
# criterion(origin_InceptionResnet_model_2, mediate_InceptionResnet_model_2) + \
# criterion(origin_IR_50_model_1, mediate_IR_50_model_1) + \
# criterion(origin_IR_SE_50, mediate_IR_SE_50) + \
# criterion(origin_mobileFaceNet, mediate_mobileFaceNet) + \
# criterion(origin_Insightface_iresent34, mediate_Insightface_iresent34) + \
# criterion(origin_Insightface_iresent50, mediate_Insightface_iresent50) + \
# criterion(origin_Insightface_iresent100, mediate_Insightface_iresent100) + \
# criterion(origin_VGG16, mediate_VGG16)
loss1 = criterion(origin_InceptionResnet_model_1, mediate_InceptionResnet_model_1) * 0.7 + \
criterion(origin_InceptionResnet_model_2, mediate_InceptionResnet_model_2) * 0.7 + \
criterion(origin_IR_50_model_1, mediate_IR_50_model_1) * 0.8 + \
criterion(origin_IR_152_model_1, mediate_IR_152_model_1) * 0.8 + \
criterion(origin_IR_SE_50, mediate_IR_SE_50) * 0.7 + \
criterion(origin_mobileFaceNet, mediate_mobileFaceNet) * 0.7 + \
criterion(origin_Insightface_iresent34, mediate_Insightface_iresent34) * 0.8 + \
criterion(origin_Insightface_iresent50, mediate_Insightface_iresent50) * 0.9 + \
criterion(origin_Insightface_iresent100, mediate_Insightface_iresent100) * 0.9 + \
criterion(origin_VGG16, mediate_VGG16) * 0.7
loss2 = criterion(mediate_InceptionResnet_model_1, mean_face) *0.7+ \
criterion(mediate_InceptionResnet_model_2, mean_face) *0.7+ \
criterion(mediate_IR_50_model_1, mean_face) *0.8+ \
criterion(mediate_IR_152_model_1, mean_face) *0.8+\
criterion(mediate_IR_SE_50, mean_face) *0.7+ \
criterion(mediate_mobileFaceNet, mean_face) *0.7+ \
criterion(mediate_Insightface_iresent34, mean_face) *0.8+ \
criterion(mediate_Insightface_iresent50, mean_face) *0.9+ \
criterion(mediate_Insightface_iresent100, mean_face)*0.9 + \
criterion(mediate_VGG16, mean_face)*0.7
loss3 = criterion(average_out,mediate_InceptionResnet_model_1)+ \
criterion(average_out,mediate_InceptionResnet_model_2)+ \
criterion(average_out,mediate_IR_50_model_1) + \
criterion(average_out,mediate_IR_152_model_1) + \
criterion(average_out,mediate_mobileFaceNet) + \
criterion(average_out,mediate_Insightface_iresent34)+ \
criterion(average_out,mediate_Insightface_iresent50) + \
criterion(average_out,mediate_Insightface_iresent100)+ \
criterion(average_out,mediate_VGG16)+ \
criterion(average_out,mediate_IR_SE_50)
loss = alpha * loss1 - beta * loss2 - gamma * loss3
print('loss : %f' % loss)
print('loss1 : %f' % loss1)
print('loss2 : %f' % loss2)
# compute gradients
loss.backward(retain_graph=True)
g_noise = momentum * g_noise + (
in_variable.grad / in_variable.grad.data.norm(1)) * 0.9
g_noise = g_noise / g_noise.data.norm(1)
g1 = g_noise
g2 = g_noise
# if i % 3 == 0 :
kernel = gkern(3, 2).astype(np.float32)
gaussian_blur1 = GaussianBlur(kernel)
gaussian_blur1.to(device)
g1 = gaussian_blur1(g1)
g1 = torch.clamp(g1, -0.2, 0.2)
# else:
addition = TVLoss()
addition.to(device)
g2 = addition(g2)
g_noise = 0.25 * g1 + 0.75 * g2
g_noise.clamp_(-0.05, 0.05)
in_variable.data = in_variable.data + (
(eps / 255.) * torch.sign(g_noise)
) # * torch.from_numpy(mat).unsqueeze(0).float()
in_variable.grad.data.zero_() # unnecessary
# g_noise = in_variable.data - origin_variable
# g_noise.clamp_(-0.2, 0.2)
# in_variable.data = origin_variable + g_noise
# deprocess image
adv = in_variable.data.cpu().numpy()[0] # (3, 112, 112)
perturbation = (adv - in_tensor.cpu().numpy())
adv = adv * 128.0 + 127.0
adv = adv.swapaxes(0, 1).swapaxes(1, 2)
adv = adv[..., ::-1]
adv = np.clip(adv, 0, 255).astype(np.uint8)
# sample_dir = './target_mean_face/'
# if not os.path.exists(sample_dir):
# os.makedirs(sample_dir)
advimg = sample_dir + path.split('/')[-1].split('.')[-2] + '.jpg'
cv2.imwrite(advimg, adv)
print("save path is " + advimg)
print('cost time is %.2f s ' % (time.time() - start))
